Week4__Neural_Networks.ipynb

{
  "cells": [
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "view-in-github",
        "colab_type": "text"
      },
      "source": [
        "<a href=\"https://colab.research.google.com/github/IAT-ExploringAI-2024/Week4-NeuralNetworks/blob/main/Week4__Neural_Networks.ipynb\" target=\"_parent\"><img src=\"https://colab.research.google.com/assets/colab-badge.svg\" alt=\"Open In Colab\"/></a>"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "hKtwydx6Z7BR",
        "outputId": "93b60029-152f-458a-bbaa-b09700d0e7e6"
      },
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Mounted at /content/drive\n"
          ]
        }
      ],
      "source": [
        "from google.colab import drive\n",
        "drive.mount('/content/drive')"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "q5CvOjjrL9gw"
      },
      "source": [
        "<center><h1> Introduction to Audio Classification with Neural Networks</h1></center>\n",
        "\n",
        "# Abstract\n",
        "\n",
        "### Purpose\n",
        "This notebook serves as an introduction to working with audio data for classification problems; it is meant as a learning resource rather than a demonstration of the state-of-the-art. The techniques mentioned in this notebook apply not only to classification problems, but to regression problems and problems dealing with other types of input data as well. I focus particularly on feature engineering techniques for audio data and provide an in-depth look at the logic, concepts, and properties of the Multilayer Perceptron (MLP) model, an ancestor and the origin of deep neural networks (DNNs) today. I also provide an introduction to a few key machine learning models and the logic in choosing their hyperparameters. These objectives are framed by the task of recognizing emotion from snippets of speech audio.\n",
        "\n",
        "### Summary\n",
        "Data cleansing and feature engineering comprise the most crucial aspect of preparing machine and deep learning models alike and is often the difference between success and failure. We can drastically improve the performance of a model with proper attention paid to feature engineering. This stands for input data which is already useable for predictions; even such data can be transformed in myriad ways to improve predictive performance. For features to be useful in classification they must encompass sufficient variance between different classes. We can further improve the performance of our models by understanding the influence of and precisely tuning their hyperparameters, for which there are algorithmic aids such as Grid Search.\n",
        "\n",
        "Network architecture is a critical factor in determining the computational complexity of DNNs; often, however, simpler models with just one hidden layer perform better than more complicated models. The importance of proper model evaluation cannot be overstressed: training data should be used strictly for training a model, validation data strictly for tuning a model, and test data strictly to evaluate a model once it is tuned - a model should never be tuned to perform better on test data. To this end, K-Fold Cross Validation is a staple tool.\n",
        "\n",
        "### Conclusions\n",
        "Classic machine learning models such as Support Vector Machines (SVM), k Nearest Neighbours (kNN), and Random Forests have distinct advantages to deep neural networks in many tasks but do not match the performance of even the simplest deep neural network in the task of audio classification. The Multilayer Perceptron (MLP) model is the simplest form of DNN suited to classification tasks, provides decent off-the-shelf performance, and can be precisely tuned to be accurate and relatively quick to train.\n",
        "\n",
        "The MLP provides appreciable accuracy on the RAVDESS dataset, but suffers from the relatively small number of training samples afforded by this dataset. Long Short Term Memory Recurrent Neural Networks (LSTM RNNs) and Convolutional Neural Networks (CNNs) are excellent DNN candidates for audio data classification: LSTM RNNs because of their excellent ability to interpret sequential data such as the audio waveform represented as a time series, and CNNs because features engineered on audio data such as spectrograms have marked resemblance to images, in which CNNs excel at recognition and discrimination between distinct patterns.\n",
        "\n"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "QQsTfGREL9g1"
      },
      "source": [
        "<!--TABLE OF CONTENTS-->\n",
        "\n",
        "\n",
        "# Table of Contents\n",
        "  - [Intro: Speech Emotion Recognition on the RAVDESS dataset](#Intro:-Speech-Emotion-Recognition-on-the-RAVDESS-dataset)\n",
        "  - [Machine Learning Process Overview](#Machine-Learning-Process-Overview)\n",
        "  - [Feature Extraction](#Feature-Extraction)\n",
        "    - [Load the Dataset and Compute Features](#Load-the-Dataset-and-Compute-Features)\n",
        "    - [Feature Scaling](#Feature-Scaling)\n",
        "  - [The MLP Model for Classification](#The-MLP-Model-for-Classification)\n",
        "    - [Choice of Hyperparameters](#Choice-of-Hyperparameters)\n",
        "    - [Network Architecture](#Network-Architecture)\n",
        "    - [Hyperparameter Optimization and Grid Search](#Hyperparameter-Optimization-and-Grid-Search)\n",
        "  - [Training and Evaluating the MLP Model](#Training-and-Evaluating-the-MLP-Model)\n",
        "    - [The Confusion Matrix](#The-Confusion-Matrix)\n",
        "    - [Precision, Recall, F-Score](#Precision,-Recall,-F-Score)\n",
        "    - [K-Fold Cross-Validation](#K-Fold-Cross-Validation)\n",
        "    - [The Validation Curve: Further Tuning of Hyperparameters](#The-Validation-Curve:-Further-Tuning-of-Hyperparameters)\n",
        "    - [The Learning Curve: Determining Optimal Training Set Size](#The-Learning-Curve:-Determining-Optimal-Training-Set-Size)\n",
        "  - [Higher Complexity DNNs](#Higher-Complexity-DNNs)"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "hZUcbr4PL9g2"
      },
      "source": [
        "## Intro: Speech Emotion Recognition on the RAVDESS dataset\n",
        "In this notebook, we train a Multilayer Perceptron (MLP) model for classification in an attempt to recognize the emotion conveyed in a speech audio snippet. MLP classifiers are a good DNN model to start with because they are simple, flexible, and suited when inputs are assigned a label - in our case, emotion.\n",
        "\n",
        "We're going to use the RAVDESS dataset (Ryerson Audio-Visual Database of Emotional Speech and Song dataset), created by Steven Livingstone and Frank Russo of Ryerson University. <br>\n",
        "[Details of the RAVDESS dataset](https://smartlaboratory.org/ravdess/) <br>\n",
        "[Download the dataset used in this notebook](https://1sfu-my.sharepoint.com/:f:/g/personal/oyalcin_sfu_ca/ErKIxg5g4rFOlfrAZ352DW4BD1ytBiz1kZLcj5Elk9_1rQ?e=lgUQoi) <br> Scroll half-way down the page and find \"Audio_Speech_Actors_01-24\"<br>\n",
        "\n",
        "We're going to use the audio-only speech portion of the RAVDESS dataset, ~200MB.\n",
        "Audio is sourced from 24 actors (12 male, 12 female) repeating two sentences with\n",
        "a variety of emotions and intensity. We get 1440 speech files (24 actors * 60 recordings per actor). Each audio sample has been rated  by a human 10 times for emotional quality.\n",
        "\n",
        "## Machine Learning Process Overview\n",
        "1. Feature Engineering: Choose and define the properties which our model will use to evaluate the audio files. <br>\n",
        "2. Feature Extraction: Compute the features for each audio file and build a feature matrix representing all audio files. <br>\n",
        "3. Model exploration: Test candidate models that make sense for the properies of the dataset\n",
        "4. Training the MLP Classifier model: Choose and optimize the properties of our model on validation data - hyperparameters and architechture.  <br>\n",
        "5. Evaluate our model's performance: Evaluate our model's accuracy on validation data and score it against test data which it has never seen in training.<br>\n",
        "6. Explore options for improving our model: Is our dataset the right size? Is our model too complex or too simple? <br>\n"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 56,
      "metadata": {
        "id": "Zg6HC4zIaZKm"
      },
      "outputs": [],
      "source": [
        "#importing the required libraries\n",
        "\n",
        "import matplotlib.pyplot as plt\n",
        "import pandas as pd\n",
        "import numpy as np\n",
        "import librosa.display\n",
        "import soundfile\n",
        "import os\n",
        "# matplotlib complains about the behaviour of librosa.display, so we'll ignore those warnings:\n",
        "import warnings; warnings.filterwarnings('ignore')"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "eu8VPhDmL9hC"
      },
      "source": [
        "## Feature Extraction\n",
        "We're going to repeat the feature extraction process from previous weeks, and calculate the following:\n",
        "\n",
        "**Chromagram**: Will produce 12 features; One for each of 12 pitch classes\n",
        "\n",
        "**Mel Spectrogram**: Will produce 128 features; We've defined the number of mel frequency bands at n_mels=128\n",
        "\n",
        "**MFCC**: Will produce 40 MFCCs; I've set the number of coefficients to return at n_mfcc=40 which I found to work well"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 57,
      "metadata": {
        "id": "qTe93WYTL9hD"
      },
      "outputs": [],
      "source": [
        "\n",
        "\n",
        "def feature_chromagram(waveform, sample_rate):\n",
        "    # STFT computed here explicitly; mel spectrogram and MFCC functions do this under the hood\n",
        "    stft_spectrogram=np.abs(librosa.stft(waveform))\n",
        "    # Produce the chromagram for all STFT frames and get the mean of each column of the resulting matrix to create a feature array\n",
        "    chromagram=np.mean(librosa.feature.chroma_stft(S=stft_spectrogram, sr=sample_rate).T,axis=0)\n",
        "    return chromagram\n",
        "\n",
        "def feature_melspectrogram(waveform, sample_rate):\n",
        "    # Produce the mel spectrogram for all STFT frames and get the mean of each column of the resulting matrix to create a feature array\n",
        "    # Using 8khz as upper frequency bound should be enough for most speech classification tasks\n",
        "    melspectrogram=np.mean(librosa.feature.melspectrogram(y=waveform, sr=sample_rate, n_mels=128, fmax=8000).T,axis=0)\n",
        "    return melspectrogram\n",
        "\n",
        "def feature_mfcc(waveform, sample_rate):\n",
        "    # Compute the MFCCs for all STFT frames and get the mean of each column of the resulting matrix to create a feature array\n",
        "    # 40 filterbanks = 40 coefficients\n",
        "    mfc_coefficients=np.mean(librosa.feature.mfcc(y=waveform, sr=sample_rate, n_mfcc=40).T, axis=0)\n",
        "    return mfc_coefficients"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "xjFXIfC2L9hD"
      },
      "source": [
        "We're going to wrap our feature extraction functions so we only have to load each audio file once. After extracting our 3 audio features as NumPy arrays representing a time series, we're going to\n",
        "stack them horizontally to create a single feature array."
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "def preprocess_waveform(waveform):\n",
        "    # If the waveform has 2 channels (stereo), convert it to mono\n",
        "    if len(waveform.shape) > 1:\n",
        "        waveform = librosa.to_mono(waveform)\n",
        "    return waveform"
      ],
      "metadata": {
        "id": "jTQHdf5Dpl0g"
      },
      "execution_count": 59,
      "outputs": []
    },
    {
      "cell_type": "code",
      "execution_count": 60,
      "metadata": {
        "id": "xPMw9ijJL9hE"
      },
      "outputs": [],
      "source": [
        "def get_features(file):\n",
        "    # load an individual soundfile\n",
        "     with soundfile.SoundFile(file) as audio:\n",
        "        waveform = audio.read(dtype=\"float32\")\n",
        "        sample_rate = audio.samplerate\n",
        "        # make sure the file is mono channel audio\n",
        "        waveform = preprocess_waveform(waveform)\n",
        "        # compute features of soundfile\n",
        "        chromagram = feature_chromagram(waveform, sample_rate)\n",
        "        melspectrogram = feature_melspectrogram(waveform, sample_rate)\n",
        "        mfc_coefficients = feature_mfcc(waveform, sample_rate)\n",
        "\n",
        "        feature_matrix=np.array([])\n",
        "        # use np.hstack to stack our feature arrays horizontally to create a feature matrix\n",
        "        feature_matrix = np.hstack((chromagram, melspectrogram, mfc_coefficients))\n",
        "\n",
        "        return feature_matrix"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "c-C6g6psL9hE"
      },
      "source": [
        "### Load the Dataset and Compute Features\n",
        "We have to understand the labelling of the RAVDESS dataset to find the ground truth emotion for each sample.\n",
        "Each file is labelled with 7 numbers delimited by a \"-\".\n",
        "Most of the numbers describe metadata about the audio samples such as their format (video and/or audio),\n",
        "whether the audio is a song or statement, which of two statements is being read and by which actor.\n",
        "\n",
        "The third and fourth numbers pertain to the emotional quality of each sample. The third number is in the range of 1-8 with each number representing an emotion.\n",
        "The fourth number is either 1 or 2, representing normal (1) or strong (2) emotional intensity.\n",
        "\n",
        "We're going to define a dictionary based on the third number (emotion) and assign an emotion to each number as specified by the RAVDESS dataset:"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 61,
      "metadata": {
        "id": "z-Pu_fB7L9hF"
      },
      "outputs": [],
      "source": [
        "#Emotions in the RAVDESS dataset\n",
        "emotions_dict ={\n",
        "  '01':'neutral',\n",
        "  '02':'calm',\n",
        "  '03':'happy',\n",
        "  '04':'sad',\n",
        "  '05':'angry',\n",
        "  '06':'fearful',\n",
        "  '07':'disgust',\n",
        "  '08':'surprised'\n",
        "}"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "pTlsUOwXL9hF"
      },
      "source": [
        "Finally, let's load our entire dataset and compute the features of each audio file:"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 66,
      "metadata": {
        "id": "mAh2AYMpL9hF"
      },
      "outputs": [],
      "source": [
        "import os, glob\n",
        "\n",
        "def load_data():\n",
        "    X,y=[],[]\n",
        "    count = 0\n",
        "    for file in glob.glob(\"Audio Data/Audio Data/*/*.wav\"):\n",
        "        file_name=os.path.basename(file)\n",
        "        emotion=emotions_dict[file_name.split(\"-\")[2]]\n",
        "        features = get_features(file)\n",
        "        X.append(features)\n",
        "        y.append(emotion)\n",
        "        count += 1\n",
        "        # '\\r' + end='' results in printing over same line\n",
        "        print('\\r' + f' Processed {count}/{1440} audio samples',end=' ')\n",
        "    # Return arrays to plug into sklearn's cross-validation algorithms\n",
        "    return np.array(X), np.array(y)"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "3Cf8q4K5L9hG"
      },
      "source": [
        "Compute the feature matrix and read the emotion labels for the entire dataset.\n",
        "Note that our regressor (independent/explanatory variable), usually denoted X, is named 'features', and our regressand (dependent variable), usually denoted y, is named 'emotions'."
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 67,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "XeKE591aL9hG",
        "outputId": "f1319292-751b-4e71-f721-a3dd52f0abda"
      },
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            " Processed 1439/1440 audio samples "
          ]
        }
      ],
      "source": [
        "features, emotions = load_data()"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "MzE6ZwqxoRVM"
      },
      "source": [
        "We're going to create dataframes of both features and emotions, and save them into .csv, not to do this again. You can also use .csv's from prior weeks."
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 68,
      "metadata": {
        "id": "2tE2z85ioRVS",
        "outputId": "11d12ca0-6d65-498e-f9e8-28c32ee5e044",
        "colab": {
          "base_uri": "https://localhost:8080/"
        }
      },
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "\n",
            "Audio samples represented: 1439\n",
            "Numerical features extracted per sample: 180\n"
          ]
        }
      ],
      "source": [
        "print(f'\\nAudio samples represented: {features.shape[0]}')\n",
        "print(f'Numerical features extracted per sample: {features.shape[1]}')\n",
        "features_df = pd.DataFrame(features) # make it pretty for display\n",
        "features_df.to_csv('featuresRavdess.csv')\n",
        "\n",
        "#making dataframe for emotions as well\n",
        "emotions_df = pd.DataFrame(emotions) # make it pretty for display\n",
        "emotions_df.to_csv('emotionsRavdess.csv')\n"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "VpE5m-5aEyoB"
      },
      "source": [
        "## Load pre-saved Dataset"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "hHPB7dCqEotR"
      },
      "source": [
        "Once saved you only need to load them later by running the cell below, and **skip every cell above** except for the one in which we import libraries."
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 69,
      "metadata": {
        "id": "-8nlJCESEn56"
      },
      "outputs": [],
      "source": [
        "features=pd.read_csv('featuresRavdess.csv',index_col=0)\n",
        "emotions=pd.read_csv('emotionsRavdess.csv',index_col=0)"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "RVUa7RAuL9hG"
      },
      "source": [
        "Let's see what the features we extracted look like:"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 70,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "mzxX583yL9hG",
        "outputId": "9c112748-16da-44bd-ecf4-61f8469a374a"
      },
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "\n",
            "Audio samples represented: 1439\n",
            "Numerical features extracted per sample: 180\n"
          ]
        }
      ],
      "source": [
        "print(f'\\nAudio samples represented: {features.shape[0]}')\n",
        "print(f'Numerical features extracted per sample: {features.shape[1]}')\n",
        "features_df = pd.DataFrame(features) # make it pretty for display\n",
        "labels_df=pd.DataFrame(emotions)"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "Tq083zLEL9hH"
      },
      "source": [
        "We have a matrix of dim 1435 x 180. Looks good - 1435 audio samples, one per row, with a series of\n",
        "180 numerical features for each sample.\n",
        "\n",
        "**Each of the 1435 feature arrays has 180 features composed of 12 chromagram pitch classes + 128 mel spectrogram bands + 40 MFC coefficients.**"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "WhtlYshOL9hV"
      },
      "source": [
        "### Feature Scaling\n",
        "We're going to also do feature scaling, similar to what we did in the last weeks."
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 72,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "_IUm1DLwL9hW",
        "outputId": "155a6300-848f-48cf-a0db-0a8443596c29"
      },
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "12 Chromagram features:           min = 0.310,     max = 1.000,     mean = 0.667,     deviation = 0.088\n",
            "\n",
            "128 Mel Spectrogram features:     min = 0.000,     max = 149.208,     mean = 0.188,     deviation = 1.604\n",
            "\n",
            "40 MFCC features:                 min = -1131.371,    max = 115.126,    mean = -15.073,    deviation = 100.101\n"
          ]
        }
      ],
      "source": [
        "# We would usually use df.describe(), but it provides a bit of a mess of information we don't need at the moment.\n",
        "def print_features(df):\n",
        "    # Check chromagram feature values\n",
        "    features_df_chromagram = df.iloc[:,:11]\n",
        "    chroma_min = features_df_chromagram.min().min()\n",
        "    chroma_max = features_df_chromagram.max().max()\n",
        "    # stack all features into a single series so we don't get a mean of means or stdev of stdevs\n",
        "    chroma_mean = features_df_chromagram.stack().mean()\n",
        "    chroma_stdev = features_df_chromagram.stack().std()\n",
        "    print(f'12 Chromagram features:       \\\n",
        "    min = {chroma_min:.3f}, \\\n",
        "    max = {chroma_max:.3f}, \\\n",
        "    mean = {chroma_mean:.3f}, \\\n",
        "    deviation = {chroma_stdev:.3f}')\n",
        "\n",
        "    # Check mel spectrogram feature values\n",
        "    features_df_melspectrogram = df.iloc[:,12:139]\n",
        "    mel_min = features_df_melspectrogram.min().min()\n",
        "    mel_max = features_df_melspectrogram.max().max()\n",
        "    # stack all features into a single series so we don't get a mean of means or stdev of stdevs\n",
        "    mel_mean = features_df_melspectrogram.stack().mean()\n",
        "    mel_stdev = features_df_melspectrogram.stack().std()\n",
        "    print(f'\\n128 Mel Spectrogram features: \\\n",
        "    min = {mel_min:.3f}, \\\n",
        "    max = {mel_max:.3f}, \\\n",
        "    mean = {mel_mean:.3f}, \\\n",
        "    deviation = {mel_stdev:.3f}')\n",
        "\n",
        "    # Check MFCC feature values\n",
        "    features_df_mfcc = df.iloc[:,140:179]\n",
        "    mfcc_min = features_df_mfcc.min().min()\n",
        "    mfcc_max = features_df_mfcc.max().max()\n",
        "    # stack all features into a single series so we don't get a mean of means or stdev of stdevs\n",
        "    mfcc_mean = features_df_mfcc.stack().mean()\n",
        "    mfcc_stdev = features_df_mfcc.stack().std()\n",
        "    print(f'\\n40 MFCC features:             \\\n",
        "    min = {mfcc_min:.3f},\\\n",
        "    max = {mfcc_max:.3f},\\\n",
        "    mean = {mfcc_mean:.3f},\\\n",
        "    deviation = {mfcc_stdev:.3f}')\n",
        "\n",
        "print_features(features_df)"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "gFgQlYSlL9hW"
      },
      "source": [
        "**There's an obvious imbalance in the variance our features; Our features indeed belong to very different distributions:** our MFC coefficients' deviation is greater than the other features by orders of magnitude. That does not mean MFC coefficients are the most important feature, but rather it is a property of the way they are computed. We will certainly need to scale this feature set.\n",
        "\n",
        "We have the choice of sklearn's StandardScaler and MinMaxScaler.\n",
        "In practice, **MinMax scaling is especially useful when we know our features should be in a bounded interval**, such as pixel values in [0,255], while **standard scaling is perhaps more practical for features with unknown distributions** because centering the features at zero-mean with a standard deviation of 1 means extreme values will have less of an impact on the model's learned weights, i.e. the model is less sensitive to outliers.\n",
        "\n",
        "We'll create MinMax scaled features as well so we can give them a try later on to confirm that standard scaling is better in the absence of knowledge on the appropriate distribution for a dataset's features."
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 73,
      "metadata": {
        "id": "_BCAYVEUL9hW"
      },
      "outputs": [],
      "source": [
        "from sklearn.preprocessing import StandardScaler\n",
        "from sklearn.preprocessing import MinMaxScaler\n",
        "\n",
        "scaler = StandardScaler()\n",
        "# keep our unscaled features just in case we need to process them alternatively\n",
        "features_scaled = features\n",
        "features_scaled = scaler.fit_transform(features_scaled)\n",
        "\n",
        "scaler = MinMaxScaler()\n",
        "# keep our unscaled features just in case we need to process them alternatively\n",
        "features_minmax = features\n",
        "features_minmax = scaler.fit_transform(features_minmax)"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "r4ATx5oNL9hX"
      },
      "source": [
        "Make sure our features are properly scaled:"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 74,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "mlRuHQkKL9hX",
        "outputId": "342efc3e-4fef-4ecf-9678-568803acf52f"
      },
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "\u001b[1mStandard Scaling:\n",
            "\u001b[0m\n",
            "12 Chromagram features:           min = -3.896,     max = 4.368,     mean = -0.000,     deviation = 1.000\n",
            "\n",
            "128 Mel Spectrogram features:     min = -0.474,     max = 36.531,     mean = -0.000,     deviation = 1.000\n",
            "\n",
            "40 MFCC features:                 min = -4.803,    max = 6.238,    mean = 0.000,    deviation = 1.000\n",
            "\n",
            "\n",
            "\u001b[1mMinMax Scaling:\n",
            "\u001b[0m\n",
            "12 Chromagram features:           min = 0.000,     max = 1.000,     mean = 0.478,     deviation = 0.145\n",
            "\n",
            "128 Mel Spectrogram features:     min = 0.000,     max = 1.000,     mean = 0.014,     deviation = 0.060\n",
            "\n",
            "40 MFCC features:                 min = 0.000,    max = 1.000,    mean = 0.413,    deviation = 0.170\n"
          ]
        }
      ],
      "source": [
        "print('\\033[1m'+'Standard Scaling:\\n'+'\\033[0m')\n",
        "features_scaled_df = pd.DataFrame(features_scaled)\n",
        "print_features(features_scaled_df)\n",
        "\n",
        "print('\\n\\n\\033[1m'+'MinMax Scaling:\\n'+'\\033[0m')\n",
        "features_minmax_df = pd.DataFrame(features_minmax)\n",
        "print_features(features_minmax_df)"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "vZfzIh7DL9hX"
      },
      "source": [
        "Perfect. Zero mean and unit variance for standard scaling and in the range [0,1] for MinMax scaling - a default when we don't specify values. We can now move on to building predictive models for these features."
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "GUhPmqKbL9hi"
      },
      "source": [
        "## The MLP Model for Classification\n",
        "We're going to first try a Multilayer Perceptron (MLP) Classifier, a simple artificial neural network (ANN) model well-suited for predictions trained on labelled inputs. Note that an MLP model can be trained for regression just as well. An MLP network consists of an input layer, _n_ hidden layers, and an output layer. The logic of the network depends on its weights: An array [w<sub>0</sub>-w<sub>i</sub>] for _each node_ in all layers, one weight value for each node-node connection (edge) in the graph representation (a matrix _W_ for the network). We also have biases, an array [b<sub>0</sub>-b<sub>j</sub>] for _each layer_, one value for each node (a matrix _B_ for the network). I'll speak more to its architechture below."
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "1ShGGg4IL9hj"
      },
      "source": [
        "<img src=\"https://github.com/IAT-ExploringAI-2024/Week4-NeuralNetworks/blob/main/mlp.png?raw=true\" width=500 height=500 />\n",
        "\n",
        "\n"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "Vl2GcFlIA9Lv"
      },
      "source": [
        "###Training: The 80/20 Split and Validation\n",
        "In order to compare models, we'll have to evaluate their performance. The simplest method to do so is to train a model on a portion of our dataset and test it on the remainder. We'll use sklearn's train_test_split to create a standard 80/20 train/test split. The model is fit on 80% of\n",
        "the data and tested for performance against 20% of the data, which it has never seen in training - also called the hold-out set.\n",
        "\n",
        "More accurately, the proper modality for training and scoring a model is to\n",
        "1. Fit/train our model on a _training_ set,\n",
        "2. Evaluate the model on a _validation_ set to tune the hyperparameters for better performance,\n",
        "3. Finally score our model's true performance - its **generalizability** - against a _test_ set, aka the hold-out set.\n",
        "4. Repeat from 2. **Do not tune the model to score well on the test set**.\n",
        "\n",
        "Different set ratios are used in this approach - a usual example is 60/20/20 train/validation/test.\n",
        "\n",
        "Last week, we skipped the validation test split for simplicity. This week, we're going to include it for performing hyperparameter tuning.\n",
        "\n",
        "We will basically first do the 80/20 train/test, like we did before, and then further divide the 80 train set into 60/20 train and validation set. We'll do it for both scaled and unscaled versions:"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 78,
      "metadata": {
        "id": "erVlA8IQA9MA"
      },
      "outputs": [],
      "source": [
        "from sklearn.model_selection import train_test_split\n",
        "############# Unscaled test/train set #############\n",
        "X_train80, X_test, y_train80, y_test = train_test_split(\n",
        "    features,\n",
        "    emotions,\n",
        "    test_size=0.2,\n",
        "    random_state=42\n",
        ")\n",
        "\n",
        "# Then further divide the train set into validation\n",
        "# Note that here, to get 60%-20%, we'll need to divide the 80% train set from above to 75-25 split\n",
        "# 0.8 * 0.25 = 0.2\n",
        "X_train, X_validation, y_train, y_validation = train_test_split(\n",
        "    X_train80,\n",
        "    y_train80,\n",
        "    test_size=0.25,\n",
        "    random_state=42\n",
        ")\n",
        "\n",
        "############ Standard Scaled test/train set ###########\n",
        "# The labels/classes (y_train, y_test) never change, keep old values\n",
        "X_train_scaled80, X_test_scaled, _, _ = train_test_split(\n",
        "    features_scaled,\n",
        "    emotions,\n",
        "    test_size=0.2,\n",
        "    random_state=42\n",
        ")\n",
        "\n",
        "# Repeat to get validation, same as above\n",
        "X_train_scaled, X_validation_scaled, _, _ = train_test_split(\n",
        "    X_train_scaled80,\n",
        "    y_train80,\n",
        "    test_size=0.25,\n",
        "    random_state=42\n",
        ")\n",
        "\n",
        "############# MinMax Scaled test/train set ###############\n",
        "# The labels/classes (y_train, y_test) never change, keep old values\n",
        "X_train_minmax80, X_test_minmax, _, _ = train_test_split(\n",
        "    features_minmax,\n",
        "    emotions,\n",
        "    test_size=0.2,\n",
        "    random_state=42\n",
        ")\n",
        "\n",
        "# And get validation set\n",
        "X_train_minmax, X_validation_minmax, _, _ = train_test_split(\n",
        "    X_train_minmax80,\n",
        "    y_train80,\n",
        "    test_size=0.25,\n",
        "    random_state=42\n",
        ")"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "Hk83NEWf_-_M"
      },
      "source": [
        "### MLP Initialization"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "k8wqduoWL9hj"
      },
      "source": [
        "We initialize an MLP classification model with random weights and biases at zero, the standard modality. There are techniques for determining optimal weight initialization; the initial distribution of weights is incrediby important because they impact the direction of the gradient and will determine how fast and whether at all a model's loss function will converge.\n",
        "\n",
        "We'll try the off-the-shelf MLP model that comes with sklearn. Using default settings for neural networks is not a good idea in general - and we're going to be using the validation test to check model performance:"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 79,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "FtpfgH_eL9hj",
        "outputId": "549ea062-c5d1-447a-99bf-61e627632991"
      },
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Possible emotions predicted by model:['angry' 'calm' 'disgust' 'fearful' 'happy' 'neutral' 'sad' 'surprised']\n",
            "Unscaled MLP Model's accuracy on training set is 64.43%\n",
            "Unscaled MLP Model's accuracy on validation set is 48.61%\n"
          ]
        }
      ],
      "source": [
        "from sklearn.neural_network import MLPClassifier\n",
        "\n",
        "# Default 'off-the-shelf' MLP from sklearn\n",
        "model = MLPClassifier(\n",
        "    random_state = 42\n",
        ")\n",
        "\n",
        "# Use the split train test to train, we'll first use the unscaled parameters\n",
        "model.fit(X_train, y_train)\n",
        "\n",
        "print(f'Possible emotions predicted by model:{model.classes_}')\n",
        "print(f'Unscaled MLP Model\\'s accuracy on training set is {100*model.score(X_train, y_train):.2f}%')\n",
        "print(f'Unscaled MLP Model\\'s accuracy on validation set is {100*model.score(X_validation, y_validation):.2f}%')\n",
        "\n"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "TDfFZbEcL9hk"
      },
      "source": [
        "Great - we expected terrible performance with no thought given to the model or feature scale. Otherwise, what would a machine learning engineer do? At any rate, we at least know that our model's possible predictions - classes - are correct, and it's even doing a little better than guessing with a 1/8 (12.5%) chance. More than we can ask for with this approach.\n",
        "\n",
        "Let's see how feature scaling impacts the performance of the MLP:"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 80,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "zD6n75oBL9hk",
        "outputId": "c038f917-981c-4221-b1c3-09fbc5fbef06"
      },
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "MinMax scaled MLP Model's accuracy on training set is 64.43%\n",
            "MinMax sacled MLP Model's accuracy on validation set is 46.88%\n",
            "\n",
            "Standard scaled MLP Model's accuracy on training set is 98.61%\n",
            "Standard scaled MLP Model's accuracy on validation set is 60.76%\n"
          ]
        }
      ],
      "source": [
        "\n",
        "# Fit the model again on the minmax scaled features\n",
        "model.fit(X_train_minmax, y_train)\n",
        "\n",
        "print(f'MinMax scaled MLP Model\\'s accuracy on training set is {100*model.score(X_train_minmax, y_train):.2f}%')\n",
        "print(f'MinMax sacled MLP Model\\'s accuracy on validation set is {100*model.score(X_validation_minmax, y_validation):.2f}%\\n')\n",
        "\n",
        "\n",
        "# Fit the model again on the scaled features\n",
        "model.fit(X_train_scaled, y_train)\n",
        "\n",
        "print(f'Standard scaled MLP Model\\'s accuracy on training set is {100*model.score(X_train_scaled, y_train):.2f}%')\n",
        "print(f'Standard scaled MLP Model\\'s accuracy on validation set is {100*model.score(X_validation_scaled, y_validation):.2f}%')"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "FmgWuxwoL9hl"
      },
      "source": [
        "That's more like it. Standard scaling is indeed the way to go here. **It is important to consider choice of scaling method, if necessary, as a crucial hyperparameter of a model.** Skipping this step or choosing the wrong scaling method can render even the most appropriate features, worthless.\n",
        "\n",
        "Perhaps with some effort we can make the MLP model work for us. We'll try to give it a fighting chance by actually paying attention to what it's doing:"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "xsieghrYL9hl"
      },
      "source": [
        "### Choice of Hyperparameters\n",
        "Before training a network we must select hyperparameters, which determine the network's learning behaviour - hyperparameters determine how the network learns its weights and biases, while the network parameters determine what those weights and biases are. Each type of model (SVC, MLP, Random Forests, etc...) come with their own set of hyperparameters, and each type of optimization algorithm does as well. We may use the same optimization algorithm for two different models, but memorizing the best hyperparameters for one model won't help us with any other model and likely, not any other task even with the same model.\n",
        "\n",
        "The MLP model has a number of hyperparameters significant to its learning behaviour:\n",
        "- **Alpha**: Constrains the model's weights to be within a certain bound to address overfitting; in range [0,1], parameterizes l2 penalty which defines how steeply the cost function modifies weights in proportion to their magnitude, i.e. **regularizes** the cost function and so is also called the regularization term. In practice, **higher alpha more steeply penalizes large weights.**\n",
        "- **Activation function:** Determines the output of a neuron by the transformation applied to the set of inputs to that neuron\n",
        "- **Solver:** Algorithm(s) used for optimization of our weights (with backpropogation **gradient descent** in the case of the MLP classifier)\n",
        "- **Learning rate:** How large of a change the optimization algorithm makes to the model's weights at each training iteration, in\n",
        "- **Epsilon:** Unique to the 'adam' solver, numerical stability - to avoid divison by zero.\n",
        "\n",
        "### Network Architecture\n",
        "The input layer of our MLP neural network is the size of our feature space - that is, one neuron per feature. In our case, 180 input neurons constitute the input layer. Since we are performing multiclass classification, our MLP network has one neuron in its output layer for each class label. In our case 8 output neurons, one per emotion.\n",
        "\n",
        "The size of our hidden layer is of interest and a more dubious task than choosing other hyperparameters, because the behaviour of different numbers and sizes of hidden layers is much less transparent. There are many opinions on the matter, but there is relative consensus in that **many tasks rarely benefit from more than one hidden layer**, i.e. one hidden layer is sufficient in most cases. The number of neurons in that hidden layer is a less obvious choice, but it seems from literature that a number between the input layer and output layer size is a good starting point. At any rate, the time complexity of training our model scales exponentially with the number of hidden layers, so let's keep that number at the minimum, and likely optimum of 1.\n",
        "\n",
        "### Hyperparameter Optimization and Grid Search\n",
        "We're going to take advantage of the grid search cross-validation algorithm to find the best hyperparameters for us.\n",
        "\n",
        "Grid search is relatively simple to understand - we specify a set of candidate values for each hyperparameter, and grid search tries all combinations of those hyperparameter values to build and score potential models against our training and validation data. Grid search scores sets of hyperparameters by building a validation set from the training set we give it.\n",
        "\n",
        "We initialize the MLP classifier with the hyperparameters we will keep constant, those less likely to have a significant impact on the model's performance. We shouldn't use grid search to look for hyperparameters we can easily tune later, such as the number of training iterations (# of forward+backward passes through the network for each _batch_ of training samples); if we do so, we significantly inflate the computational cost with little return. For these hyperparameters we should choose values which are usually reasonable from a literature search (...from stackexchange).\n",
        "\n",
        "In case grid search is too expensive for the hyperparameter grid we specify, **Randomized Grid Search** can be used to sample a fixed number of hyperparameter combinations from the distribution of settings we define.\n",
        "\n",
        "We define a random state so that we can accurately compare any improvements we make to the model.\n",
        "\n",
        "We also define a **batch size**: the number of training samples included in one forward (input) and backward (error) pass of the model, which is one **iteration**. A batch size in range [1, #samples] is common - in our case, [1,1440\\*0.8]. [**Mini-batches**](https://machinelearningmastery.com/gentle-introduction-mini-batch-gradient-descent-configure-batch-size/) in the set [32, 64, 128, 256] are usually encouraged, especially on smaller datasets. **Smaller batches** cause more frequent weight updates and faster convergence - faster learning - though each update is less accurate compared to a larger batch. **Larger batches** train on a larger set of inputs simultaneously, and so may optimize loss better at each iteration due to a more accurate gradient computation - however, large batches may encourage overfitting and worsen generalization error."
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 81,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "L9fnJQHzL9hm",
        "outputId": "3c6ec297-9b66-490e-9e59-873fda1bed1b"
      },
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Fitting 10 folds for each of 48 candidates, totalling 480 fits\n",
            "Best parameters found:\n",
            " {'activation': 'relu', 'alpha': 0.001, 'hidden_layer_sizes': (300,), 'learning_rate': 'adaptive', 'solver': 'adam'}\n"
          ]
        }
      ],
      "source": [
        "from sklearn.model_selection import GridSearchCV\n",
        "from joblib import parallel_backend\n",
        "\n",
        "\n",
        "# Now that we know standard scaling is best for our features, we'll use those for our training and test sets\n",
        "X_train, X_test, y_train, y_test = train_test_split(\n",
        "    features_scaled,\n",
        "    emotions,\n",
        "    test_size=0.2,\n",
        "    random_state=42\n",
        ")\n",
        "\n",
        "\n",
        "# Initialize the MLP Classifier and choose parameters we want to keep constant\n",
        "model = MLPClassifier(\n",
        "    # tune batch size later\n",
        "    batch_size=256,\n",
        "    # keep random state constant to accurately compare subsequent models\n",
        "    random_state=42\n",
        ")\n",
        "\n",
        "# Choose the grid of hyperparameters we want to use for Grid Search to build our candidate models\n",
        "parameter_space = {\n",
        "    # A single hidden layer of size between 8 (output classes) and 180 (input features) neurons is most probable\n",
        "    # It's a bad idea at guessing the number of hidden layers to have\n",
        "    # ...but we'll give 2 and 3 hidden layers a shot to reaffirm our suspicions that 1 is best\n",
        "    'hidden_layer_sizes': [(180,), (300,),(100,50,)],\n",
        "    'activation': ['relu', 'logistic'],\n",
        "    'solver': ['sgd', 'adam'],\n",
        "    'alpha': [0.001, 0.01],\n",
        "    #'epsilon': [1e-08, 0.1],\n",
        "    'learning_rate': ['adaptive', 'constant']\n",
        "}\n",
        "\n",
        "# Create a grid search object which will store the scores and hyperparameters of all candidate models\n",
        "grid = GridSearchCV(\n",
        "    model,\n",
        "    parameter_space,\n",
        "    cv=10, # CV here shows how many fold of cross-validation we'll do\n",
        "    n_jobs=-1,\n",
        "    verbose=10)\n",
        "# Fit the models specified by the parameter grid\n",
        "# Note that, here we used the whole 80 training set because cross-validation holds out validation sets automatically\n",
        "# So we won't be using 60-20 split, GridSearchCV will automatically do it\n",
        "with parallel_backend('multiprocessing'):\n",
        "    grid.fit(X_train, y_train)\n",
        "\n",
        "\n",
        "# get the best hyperparameters from grid search object with its best_params_ attribute\n",
        "print('Best parameters found:\\n', grid.best_params_)"
      ]
    },
    {
      "cell_type": "code",
      "source": [],
      "metadata": {
        "id": "QSmmKan0eLm5"
      },
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "R2CV3cakL9hm"
      },
      "source": [
        "#### RELU Activation Function\n",
        "Grid search chose the (Rectified Linear Unit) relu function as the best activation function The function returns 0 if it receives any negative input, but for any positive value  x, it returns that value back. So it can be written as  f(x)=max(0,x). While our **hidden layer use the relu function** for its activation on each neuron, our **output layer will necessarily use the softmax function** - an exponential normalization function - in order to scale output neuron values to a probability between \\[0,1\\] for each class (emotion). Softmax does this normalization by taking the exponent of the value of each output neuron and dividing that by sum of the exponentiated output values, producing _n_ probabilities for _n_ classes. We then choose the highest probability as the predicted class for a set of input features. To be clear, **softmax activation is applied to the values of _all_ output neurons _once_ - not to each neuron.**    \n",
        "\n",
        "#### Adam Optimization Algorithm\n",
        "The solver chosen is the **Adam optimization** algorithm, a variant of **stochastic gradient descent** (SGD). Unlike SGD, which maintains a constant learning rate throughout each training iteration, Adam actually 'adapts', or varies the learning rate by taking into account the moving averages of the first and second moments (mean and variance) of the gradient at each training iteration. It makes sense that grid search chose the more sophisticated algorithm - note however it's not necessarily better than SGD for all tasks.  \n",
        "\n",
        "#### Architecture\n",
        "As expected, the ideal architechture involves just one hidden layer - though with double the neurons we expected for 180 features. Training the 2 and 3 hidden layer models was extremely computationally expensive compared to 1 hidden layer - and provided absolutely no benefit, only complexity and wasted resources. A good reinforcement of the power of simplicity, especially in deep neural networks. In the domain of model simplicity, there is also a compelling argument to be made for interpretability (to the point a model should _never_ be a black box) ...but that's for another time.\n",
        "\n",
        "We can now initialize our MLP model with the best hyperparameters for the task of speech emotion classification."
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "czvzUrltL9hn"
      },
      "source": [
        "## Training and Evaluating the MLP Model"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 82,
      "metadata": {
        "id": "y6uw8jnnL9hn",
        "outputId": "236203fe-c26b-4f86-dca3-cb282f3ffaf0",
        "colab": {
          "base_uri": "https://localhost:8080/"
        }
      },
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "MLP Model's accuracy on training set is 99.74%\n",
            "MLP Model's accuracy on test set is 69.44%\n"
          ]
        }
      ],
      "source": [
        "# Now we know best parameters from the Hyperparameter tuning\n",
        "# We can use them for the final model\n",
        "model = MLPClassifier(\n",
        "    activation='relu',\n",
        "    solver='adam',\n",
        "    alpha=0.001,\n",
        "    batch_size=256,\n",
        "    hidden_layer_sizes=(300,),\n",
        "    learning_rate='adaptive',\n",
        "    max_iter=1000, # I've found for this task, loss converges at ~1000 iterations\n",
        "    random_state=42,\n",
        ")\n",
        "\n",
        "# In the final model, we can use the whole training dataset\n",
        "model.fit(X_train, y_train)\n",
        "\n",
        "print(f'MLP Model\\'s accuracy on training set is {100*model.score(X_train, y_train):.2f}%')\n",
        "# As we're done with hyperparameter tuning, we can check accuracy on test set\n",
        "# Note that, up to this point, we did not touch the test set at all\n",
        "print(f'MLP Model\\'s accuracy on test set is {100*model.score(X_test, y_test):.2f}%')"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "NAthU_-uL9hn"
      },
      "source": [
        "That's a little better - but not great. That's expected, because most of the hyperparameters chosen by grid search are default with sklearn's MLP. At any rate, our MLP model is **overfitting the training data and not generalizing well to the test set**. With the near-perfect training accuracy it's clear our model's loss function is converging - let's see:"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 83,
      "metadata": {
        "id": "NuLsse6-L9hn",
        "outputId": "870cde19-90e2-4e87-8ccd-7ecd5a97617c",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 509
        }
      },
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "text/plain": [
              "<Figure size 640x480 with 1 Axes>"
            ],
            "image/png": "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"
          },
          "metadata": {}
        }
      ],
      "source": [
        "plt.figure()\n",
        "plt.plot([num for num in range(1,model.n_iter_ +1)], model.loss_curve_)\n",
        "plt.title('Loss Curve for MLP Model')\n",
        "plt.ylabel('Loss')\n",
        "plt.xlabel('Iteration')\n",
        "plt.show()"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "B1VQlJ7cL9ho"
      },
      "source": [
        "**Loss is indeed converging, and it appears to be a good learning rate - we want to avoid too high a learning rate since the model will bias data towards the last batches passed through it, and avoid too slow a learning rate because our model will take too long to converge.** Because of its perfect performance on the training data and poor performance on test data, we at this point suspect our model has too high a variance: It's learning to fit its weights so precisely to the training data to score well on it that its performance doesn't extend past that training data. If we had seen poor performance on both training and test sets, we would suspect our model to have a high bias.\n",
        "\n",
        "**To solve high variance, wherein our model is overfitting to noise present in the training data, we can decrease the number of input features and/or increase the size of the training set** so our model has more data to learn on and can generalize better to data it has not seen. **We can also use data regularization, using data augmentation techniques such as the addition of random noise to the audio samples.** Data regularization is different from weight regularization, but they work towards the same purpose of reducing overfitting.\n",
        "\n",
        "**To solve high bias, wherein our model is underfitting to the data, we could increase the number of input features** so that it can learn a better fit to the underlying data.\n",
        "\n",
        "**High variance makes the model's predictions inconsistent**, while **high bias makes them inaccurate** and vice versa, respectively. Ultimately, we aim to make a model that is both accurate and consistent: low variance and low bias with good performance. There is almost always a trade-off between the two."
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "4zmz85jdL9ho"
      },
      "source": [
        "### The Confusion Matrix\n",
        "A confusion matrix describes the performance of a classification model on test data. The axes of the matrix are labelled with all possible classes, in our case emotions - **the vertical levels/ columns represents predicted classes while the horizontal levels/rows represents the ground truth. The intersection of emotions on the matrix diagonal are correctly predicted labels.** All off-diagonal elements are incorrect predictions. We are literally checking to see where our model is confused, i.e. making incorrect predictions:"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 84,
      "metadata": {
        "id": "5_jYs-V_L9ho",
        "outputId": "a1a981a4-5014-47dc-92ec-68c93884f065",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 609
        }
      },
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "text/plain": [
              "<Figure size 1600x600 with 4 Axes>"
            ],
            "image/png": "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"
          },
          "metadata": {}
        }
      ],
      "source": [
        "from sklearn.metrics import confusion_matrix\n",
        "import seaborn as sn # I like seaborn's confusion matrix over sklearn's default\n",
        "\n",
        "# get predictions on test set\n",
        "test_emotion_predictions = model.predict(X_test)\n",
        "test_emotion_groundtruth = y_test\n",
        "\n",
        "# set labels for matrix axes from emotions\n",
        "emotion_list =['neutral','calm','happy','sad','angry','fearful','disgust','surprised']\n",
        "emotion_name = [emotion for emotion in emotion_list]\n",
        "\n",
        "# build confusion matrix and normalized confusion matrix\n",
        "conf_matrix = confusion_matrix(test_emotion_groundtruth, test_emotion_predictions, labels=emotion_list)\n",
        "conf_matrix_norm = confusion_matrix(test_emotion_groundtruth, test_emotion_predictions,normalize='true', labels=emotion_list)\n",
        "\n",
        "\n",
        "\n",
        "# make a confusion matrix with labels using a DataFrame\n",
        "confmatrix_df = pd.DataFrame(conf_matrix, index=emotion_name, columns=emotion_name)\n",
        "confmatrix_df_norm = pd.DataFrame(conf_matrix_norm, index=emotion_name, columns=emotion_name)\n",
        "\n",
        "# plot confusion matrices\n",
        "plt.figure(figsize=(16,6))\n",
        "sn.set(font_scale=1.8) # emotion label and title size\n",
        "plt.subplot(1,2,1)\n",
        "plt.title('Confusion Matrix')\n",
        "sn.heatmap(confmatrix_df, annot=True, annot_kws={\"size\": 18}) #annot_kws is value font\n",
        "plt.subplot(1,2,2)\n",
        "plt.title('Normalized Confusion Matrix')\n",
        "sn.heatmap(confmatrix_df_norm, annot=True, annot_kws={\"size\": 13}) #annot_kws is value font\n",
        "\n",
        "plt.show()"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "q7mCBPRSL9hp"
      },
      "source": [
        "The sum of elements in the left matrix is size of our test set, which is the length of y_test. That should be 0.2*1435 = 287. The matrix on the right shows each element as a percentage of samples in that class. Each row adds up to 100%, i.e. each row represents all samples of a particular emotion.\n",
        "\n",
        "**We can tell the model is most accurate at predicting 'calm', and least accurate at predicting 'happy'**. Based on this we might look into our features to see where they are not distinct enough between confusing classes and whether we can rethink our feature engineering."
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "w7kwQGVtL9hp"
      },
      "source": [
        "### Precision, Recall, F-Score\n",
        "The confusion matrix is an intuitive measure of the precision and recall of our model w.r.t. each class. **Precision is a measure of how many positive predictions are true positives, and recall is a measure of how many positives we actually predicted from all positive samples in the dataset: Lower precision means we have more false positives, while lower recall means we have more false negatives.** For this dataset, precision of each emotion is how often we correctly predict it  and recall is how many of that emotion we predicted out of all samples with that emotion in the dataset.\n",
        "\n",
        "**Some tasks are more concerned with maximizing precision - minimizing false positives - such as predicting whether a prisoner is likely to reoffend if paroled.** It is [catastrophic](https://www.technologyreview.com/2019/01/21/137783/algorithms-criminal-justice-ai/) to deny an inmate parole due to a false positive.\n",
        "\n",
        "**Some tasks are more concerned with maximizing recall - minimizing false negatives - such as predicting cancer risk from biomarkers.** It is better to inconvenience 10 people with a false positive test so we can catch the 1 true positive who would otherwise have been a false negative, rather than skip the 11 tests altogether (with 10 true negatives and 1 false negative).\n",
        "\n",
        "**The [F-score](https://en.wikipedia.org/wiki/F1_score) is a weighted average of precision and recall when we are equally concerned with both metrics.**\n",
        "\n",
        "**We can also make an [ROC curve](https://en.wikipedia.org/wiki/Receiver_operating_characteristic#:~:text=A%20receiver%20operating%20characteristic%20curve,why%20it%20is%20so%20named.) by plotting True Positive vs False Positive Rate to visualize their trade-off and make considerations for further tuning.** However, ROC curves are usually plotted for binary classification; doing so for"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 85,
      "metadata": {
        "id": "XjMV8LA4L9hp",
        "outputId": "1a076d55-ad25-48a3-a3c9-1a5c109318b7",
        "colab": {
          "base_uri": "https://localhost:8080/"
        }
      },
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Test Set Accuracy score =  69.444%\n",
            "Test Set Precision score =  69.093%\n",
            "Test Set Recall score =  68.329%\n",
            "Test Set F-score score =  68.1%\n"
          ]
        }
      ],
      "source": [
        "from sklearn.metrics import f1_score, precision_score, recall_score, accuracy_score\n",
        "\n",
        "y_pred = model.predict(X_test)\n",
        "print(f'Test Set Accuracy score =  {100*accuracy_score(y_test, y_pred):.3f}%') #same as model.score(X_test, y_test)\n",
        "print(f'Test Set Precision score =  {100*precision_score(y_test, y_pred, average=\"macro\"):.3f}%')\n",
        "print(f'Test Set Recall score =  {100*recall_score(y_test, y_pred, average=\"macro\"):.3f}%')\n",
        "print(f'Test Set F-score score =  {100*f1_score(y_test, y_pred, average=\"macro\"):.3}%')"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "onGqbVhcL9hq"
      },
      "source": [
        "**Pretty balanced across the board - about the same false positive rate and false negative rate.**"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "-WwFw8bPL9hq"
      },
      "source": [
        "### K-Fold Cross-Validation\n",
        "Finally, to highlight the importance of proper model validation and get a better idea of our model's performance,\n",
        "we're going to use K-Fold Cross-Validation (K-Fold CV). We split our training dataset into _K_ unique validation sets (_K_ sets/folds) - the train/validation split being determined by _K_ where each _validation set = (100/K)%_ of the entire dataset, the training set being composed of the remaining _K-1_ validation sets. The term cross-validation refers to validating the model on multiple validation sets.\n",
        "\n",
        "The terminology here may be somewhat confusing, because we usually **split our 80% _training_ set into a _training_ and _validation_ set at each iteration of K-Fold CV.** We reserve the actual _test_ set - the one we made originally with an 80/20 split of the entire dataset - for checking our model's performance _after_ we have tuned its hyperparameters.\n",
        "\n",
        "In this way, K-Fold CV will train and score _K_ different versions of our classifier.\n",
        "Note that while training sets overlap (unless otherwise defined, e.g. sklearn's GroupKFold), validation sets never overlap.\n",
        "We'll use 10-fold CV, with _K=10_ being a choice giving average scores of models with fairly low bias and moderate variance due in part to the resulting 90/10 train/validation ratio.\n",
        "\n",
        "**K-Fold CV trains our model on _K=10_ different, overlapping training folds and check its performance against _K=10_ validation folds.**\n",
        "\n",
        "Smaller datasets would theoretically be better scored with **higher _K_,** meaning we will have a larger portion of our dataset in training sets and a greater overlap of training samples between folds - this way we better represent a small dataset, have more training data, and hopefully provide a **better estimate of the model's true generalization error**, though at the cost of **higher variance because the _K_ estimates of our model's performance are correlated** (though there are exceptions). In the extreme case, and for computationally inexpensive models, **leave-one-out K-Fold CV (LOOCV)** can be used to construct training folds consisting of all samples-1, and 1 sample with in the validation fold; _each training fold differs by just 1 sample_ and this method provides adequate training data for each model validation on smaller training datasets.\n",
        "\n",
        "Sine we're training a classifier, we use **StratifiedKFold which preserves the percentage of samples in each class** (emotion) for each fold. Although we have a balanced dataset, Stratified K-Fold CV is especially important when classifying an imbalanced dataset. We also set shuffle=True to shuffle the order of sample classes in each fold to match the default behaviour of sklearn's train_test_split so we can accurately compare that to the K-Fold CV score.\n",
        "\n",
        "Choose _K_ for our K-Fold CV and train the MLP classifier on each set of train/validation folds:"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 87,
      "metadata": {
        "id": "XkJ06_ENL9hq",
        "outputId": "50306e46-7dd4-4525-d706-955acce4caa2",
        "colab": {
          "base_uri": "https://localhost:8080/"
        }
      },
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "KFold CV scores for MLP:\n",
            "66.38%\n",
            "70.43%\n",
            "62.61%\n",
            "66.96%\n",
            "65.22%\n",
            "61.74%\n",
            "67.83%\n",
            "66.09%\n",
            "66.09%\n",
            "55.65%\n",
            "Mean score of KFold CV for MLP: 64.90% ± 3.86%\n"
          ]
        }
      ],
      "source": [
        "from sklearn.model_selection import StratifiedKFold\n",
        "\n",
        "kfold = StratifiedKFold(\n",
        "    n_splits=10,\n",
        "    random_state=42,\n",
        "    shuffle=True\n",
        ")\n",
        "\n",
        "scores = []\n",
        "X_train=pd.DataFrame(X_train)\n",
        "y_train=pd.DataFrame(y_train)\n",
        "for train_indices, validation_indices in kfold.split(X_train,y_train):\n",
        "\n",
        "\n",
        "\n",
        "    train_x=X_train.iloc[train_indices]\n",
        "    train_y=y_train.iloc[train_indices]\n",
        "\n",
        "    validation_x=X_train.iloc[validation_indices]\n",
        "    validation_y=y_train.iloc[validation_indices]\n",
        "    # fit model to training fold\n",
        "    model.fit(train_x, train_y)\n",
        "    # score the model on validation fold\n",
        "    scores.append(model.score(validation_x, validation_y))\n",
        "\n",
        "print('KFold CV scores for MLP:');[print(f'{(100*score):.2f}%') for score in scores]\n",
        "print(f'Mean score of KFold CV for MLP: {100*np.mean(scores):.2f}% ± {100*np.std(scores):.2f}%')"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "Q5OSaLjyL9hq"
      },
      "source": [
        "Indeed, quite a bit worse, but a more accurate evaluation of this model. **Because we get considerably worse performance fitting our model on random subsets of our training data, we could surmise that our model's performance is inflated by overfitting** when we train and test it just once on a regular 80/20 split.\n",
        "\n",
        "Although K-Fold CV is computationally expensive, we're getting a lot more insight out of our data and that's a serious advantage when we have very few training samples. Tuning a model to just one validation set, such as in a 60/20/20 split may provide artifically inflated performance metrics which will disappoint when the model is applied to real-world data.\n",
        "\n"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "XjjwVvEZL9hr"
      },
      "source": [
        "We can use a learning curve to compare the performance of our model if we had trained it on smaller subsets of our training dataset, and thus get an idea of whether we might expect a performance increase by using more training data, or much less likely that we don't need as much data.\n",
        "\n",
        "We specify variable sizes of training sets to use for the learning curve to make one model for each size. Remember that since we're using 0.8\\*1440 = 1152 samples in our training set, that is the upper bound that the learning curve can check for us. Like the validation curve, sklearn's learning curve implements Stratified K-Fold CV to evaluate multiclass classification models, so again we specify _K = 10_.\n"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 88,
      "metadata": {
        "id": "2RaC-LS5L9hs",
        "outputId": "1f406205-bfa5-4a9f-a7b5-b7fb2af29b92",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 509
        }
      },
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "text/plain": [
              "<Figure size 640x480 with 1 Axes>"
            ],
            "image/png": "iVBORw0KGgoAAAANSUhEUgAAAmAAAAHsCAYAAAB18RHuAAAAOXRFWHRTb2Z0d2FyZQBNYXRwbG90bGliIHZlcnNpb24zLjguMywgaHR0cHM6Ly9tYXRwbG90bGliLm9yZy/H5lhTAAAACXBIWXMAAA9hAAAPYQGoP6dpAADatUlEQVR4nOzdd3gU1frA8e9sTW/03pGmVEGRJqCAUsQGioheseC1IRb06rV3ufb2U4oFEEVE6V1EWgJSpHcIEJIA6W13Z+b3x2aXhGySTbLZJOT9PI+PsHtm5uxk2bx7znveo+i6riOEEEIIIfzGUNEdEEIIIYSobiQAE0IIIYTwMwnAhBBCCCH8TAIwIYQQQgg/kwBMCCGEEMLPJAATQgghhPAzCcCEEEIIIfxMAjAhhBBCCD+TAEwIIYQQws9MFd0BUbWMHTuW6OhoAFatWkXDhg0ruEeVX//+/Tl16hTNmjVj6dKlFd0dn9A0jTVr1rBy5Up27txJXFwcOTk5hISE0KxZM3r06MHNN99MkyZNKrqrVcrixYv57rvvOHz4MDabjZo1a3LPPfcwduzYiu5aPq73tMv48eN5+umnvT7+119/ZfLkye6/jxo1ildffbXQa+zfv79U/bzsssuKfF5RFCwWC2FhYTRr1oyePXsyatQooqKiSnytzZs3c/fdd+d7bPHixbRo0cLrc+T9fAWYM2cOnTp1KnFffMF1/7t3787333/v03N/8sknfPrpp0DFvsaKJiNgQogSWbt2LYMHD+bhhx9m3rx5HDp0iIyMDBwOB8nJyWzbto0vv/ySIUOG8NZbb2G32yu6y1XCrFmzmDhxItu2bSM1NZXs7GxOnjxJSEhIRXetWMuXLy9R+8WLF5dTT0pG13VycnJITEwkOjqaDz/8kIEDB7Jy5UqfnL8kX7gSExPZsmWLT64rqgYZARNCeO2jjz7i888/B8BgMNC/f3/69etHs2bNsFgsJCYmsnHjRn755RcyMzOZMWMGBw4c4KuvvsJisVRw7yu3zz77DACz2czTTz/NFVdcQXZ2drGjOJXBiRMn2LNnD+3atSu2bUpKChs3bvRDry6oVasWX3/9tcfnbDYb586dY/HixSxcuJCMjAwee+wx5syZw+WXX16m6y5btox///vfXrfVNK1M1xNViwRgQpSz1atXV3QXfGLatGnu4Ktu3bp8/vnntG/fvkC7AQMG8K9//YsHH3yQAwcOsGHDBt5++23++9//+rvLVUZSUhJnz54FYNCgQYwbN66Ce+SdiIgIUlNT0TSNZcuWeRWArVixArvdjsViwWaz+aGXYLFYaNu2bZFt+vfvT8uWLfnggw9QVZV33nmHH374oVTXq1GjBufOnWP//v0cO3aMpk2bFnvMkiVL3H31130RFUumIIUQxTp48CD/+9//AKhduzbfffedx+DLpX79+vzf//0foaGhAMyePbvUeTzVQVZWlvvPVSmvMjIykm7dugHOERxvuAKN3r17l1u/Suv++++nfv36AGzZsoW4uLhSnWfgwIHuP3szPRsfH8/WrVuBynlfRPmQAEwIUaxPPvnEncs1adIkr5Lr69Wrx7/+9S/AmbQ/a9ascu1jVZZ36slkqloTE4MHDwbg6NGjHDhwoMi2SUlJbNq0CYAbbrih3PtWUkajkZ49ewLO/LDiXk9hGjZs6J6+9CYPbOnSpei6TtOmTYsdqROXjqr1L11cUuLi4vj+++/566+/OHXqFHa7nVq1atGtWzdGjx5N586dizw+JyeH3377jbVr17J3716SkpKw2+2Eh4fTqlUr+vXrx+23305QUFCBY12rjQYMGMB7773H22+/zfLly8nOzqZu3brccccd3HPPPUyePJlff/2VNm3a8Ntvv3Hs2DG+/fZb/vrrL+Lj4wkMDKR169bcdNNNjBw5EoOh4HeawlZBnjx5kgEDBgDO/J/+/fszb948fvvtNw4ePEhGRgZ16tShV69e/Otf/6Jx48aF3gtN01iwYAHz5s1j7969ZGVlUadOHfr168cDDzyAoij06tULgLfeeoubb77Zq58RwLlz59xJyQ0bNmTYsGFeH3vbbbdx7NgxunXrRo8ePfI95/oZWCwW/vnnn0LPMXToUA4ePEiDBg0KTOe68qOee+45+vXrx2uvvcbWrVsxmUw0btyY+++/nyeeeAJwBo4PPPBAodc5ffo0/fv3R9d1j6vy7HY7v/76K0uXLmX//v2kpKQQGhpKmzZtGDRoEDfffHOJ89zyrgZz+fTTT92PPfLIIzz66KP5nt+2bRs//fQTW7ZsIT4+HqPRSP369bn66qsZM2YMzZo183itvO/luXPn8tFHHzF//nxSUlKoXbs2gwcPLtFKRpfrr7+e119/3T0N2bp160LbrlixAofDQf369enYsWOJr+UPNWrUcP85LS2t1OcZPHgw//zzD7t37+bkyZNFjmy6RgVLEpT++eefzJs3jx07dnD27FmsVisNGzakd+/e3HXXXdSpU6fI4xMTE/n+++/5448/iI2NxWKxcPnll3PPPfe4PyuKk5yczA8//MAff/zBiRMnyMrKokaNGnTs2JGbb76Zvn37ev16qiMJwESF+Pnnn3nttdfIycnJ9/jJkyc5efIk8+fPZ9SoUbz44ouYzeYCx+/evZuHH36YM2fOFHju7NmznD17lo0bNzJ79my+++67Qj+MHA4HDzzwQL7VR8eOHSM8PLxA22XLlvHss8/mmy7KyckhOjqa6OhoFi5cWOpk86ysLO655x42b96c7/HY2Fhmz57NL7/8wqeffurxAy07O5vHH3+cP/74o8Cx33//Pb/99huvv/56ifvksm7dOlRVBaBPnz4YjUavj61Vqxbvvvtuqa/trTNnznDHHXdw/vx592N79uyhTZs2REREkJyczKJFi4oMwBYuXIiu6wDcdNNN+Z47fvw4EyZM4PDhw/keP3/+PBs2bGDDhg18++23fPbZZzRv3tx3LywPm83Gf//7X3799dcCzx06dIhDhw4xe/ZsHn30UR566KEiz/Xss8+yaNEi999PnjxJYGBgqfrl+tIUHR3NsmXLCgSMeblWPw4ePBhFUUp1vfJ27tw5958jIyNLfZ5Bgwbx3nvvAc7Pjvvuu89ju7i4OLZv3w44A7DiRsySk5N55plnWLt2bb7HbTYbe/fuZe/evXz33Xe8/PLLjBw50uM51q9fz6OPPkpGRob7sczMTNatW8e6deuKff+AczX0U089RWpqaoHXExcXx9KlS+nfvz/vvfdelVjJWxEkABN+N2/ePF544QXAOaJy1113cfnll2M0Gjl06BAzZ85k7969zJkzB7vdzltvvZXv+KSkJP71r3+RnJyM0Wjk5ptvpl+/ftSsWZOUlBT27t3LjBkzSEpK4tixY7zzzjvu/KWLrVu3Dk3T6N+/P3fffTeqqrJy5Ur3tIpLXFwcTz31FEajkfHjx9OrVy8sFgtbt27lq6++Ij09nQ0bNjBjxowif8kX5p133iExMZFOnTpx55130qxZMxISEpg1axbr16/HZrMxefJkVq5cSXBwcL5jn3jiCXfw1b59e8aNG0fTpk2Ji4tj9uzZbNq0iUmTJpW4Ty579+51/9mV71PZfPvtt+i6zvjx47n22ms5e/Yse/fupVmzZgwdOpQffviBffv2cfjw4ULrMi1YsACAJk2a0KVLF/fjiYmJjBkzhsTERMxmM7fffjt9+/YlMjKSxMREVqxYwe+//86RI0e4++67mTdvHrVr1/aq36NHj2bgwIEkJCS43zejRo3ijjvuAKBmzZqAc4Rz0qRJ7nyiBg0acM8999C+fXtUVSU6Oppvv/2W1NRUPvjgA+x2e6GB0IEDB9i3bx9dunThwQcfJCAggD/++KNEo6IXGzx4MNHR0Rw8eJAjR454DELPnz/vrnE1ZMiQUl+rPDkcDtavXw84pyOvuOKKUp+rUaNGdOjQgV27dhUZgLmmH1u2bEmrVq2KDMCys7N56KGH2LZtGwCtW7dm7NixtG7d2h1AzZo1i+zsbCZPnoymadxyyy35zrFnzx4efPBB7HY7ZrOZO++8k/79+2M2m4mOjmbq1Kl8+eWXHkfzXTZu3MiECRNQVZUaNWowZswYunXrRkBAAMeOHWPu3LlER0ezevVqHnnkEaZOnVqiL27VhQRgwq/i4+PdUztXX301n3/+eb4pws6dOzNy5EieffZZFi5cyLx58xgyZAh9+vRxt5kxYwbJyckAPPPMM9xzzz35rtG3b19GjhzJjTfeSFpaGitXrsThcHjMrdE0ja5du/LZZ5+5P3A8Db+npKQQFBTErFmz8uVodO3ale7duzN69Gh0XefXX38tVQCWmJjIiBEjePvtt/N98A0YMIAHH3yQtWvXcv78edauXZtvmmLlypWsWbMGgGuvvZZPPvnEPWLYsWNHBg8ezOuvv16mQop5C27Wq1ev1OcpT5qm8dBDDzFx4kT3Y64g+qabbnKvZlu4cCGPP/54geP379/vzvcZPnx4vudeeuklEhMTCQkJYfr06QV+KQ8YMIBBgwYxYcIEEhMTeeutt/jggw+86netWrWoVauWe7GC67GL84AWLlzoDr66dOnC119/nW9UoXv37owcOZKxY8dy6tQpPvvsM/r16+exjIKmaTRu3Jjp06cTEBAAwFVXXeVVfwtz8TTkhAkTCrRZtmwZqqrSsGFDrrjiCk6ePFmma/qarutMmTLFnXg/aNCgfD+X0hg8eDC7du1i586dnDlzhrp16xZo45p+9CYonT59ujv4GjRoEFOmTMk3Q9CzZ0+GDx/OuHHjSElJ4ZVXXqFXr175ZgBeffVV7HY7BoOBL774Il/Sf9euXRk0aBB33nknSUlJHvuQnZ3N008/jaqqtG7dmm+//TZf4dqOHTsyYsQI3nvvPb755hs2btzITz/95P5SIS6QJHzhV7NmzSIrKwuTycQ777zjMT/LZDLx8ssvuz/8vvvuu3zPJyQkULt2bWrVqsWYMWM8XqdOnTp0794dcE4TugI2T0aNGlXktz2XO+64w2OCbKdOndy5SEeOHCnVEnKr1crzzz9foB+KonD77be7/75v3758z7vuTVBQEG+99ZbH6drJkyeXaVos7zRFWaZkylthH/CXX345LVu2BMg37ZaXa/RLUZR8049Hjx5155099NBDhY6IXHvtte7pnqVLlxIfH1+q11CYqVOnAs4SBR988IHHKZ0GDRrw9ttvA85gwnWMJyNHjnQHX75Qq1YtunbtChS+GrI0eU6+knd6Lu9/e/bsYfPmzcycOZM77riDadOmARAeHs5TTz1V5uu6vgTouu7xvpw6dYodO3YAxd8XVVWZMWMG4BwZLezfe9u2bXnuuecA52df3s/PAwcOuAO4W265xeOKy+bNmxeZC/j777+TmJgIwBtvvFHorgETJ050L9a5+DNcOEkAJvzKNVXWokWLIpNEQ0ND3dNAW7ZsyVdN/a233mLdunX8+eefHj+AXFzTN0CRQZG322AUlZiaN0E+b8Dirfbt2xMREeHxuUaNGnk8d2pqqnvpev/+/QsNjkwmU4FpiJLIGxRW1vpEderU8Ti64DJixAjAmct1ccK/ruvuwKxbt275kqXXrl3rzgu75ppriuyDa5RW07R828mUVWJiojvw7tevX5Gvs3v37u5gc/369YUW9iyPrV9cwcbevXs5ceJEvufOnj1LTEwMUDEBWGJiIjfddFOB/0aOHMndd9/Nq6++6g5M6tatyzfffEODBg3KfN1GjRq5y7V4CsBcQWmbNm2K/ZK0a9cu9xfJYcOGFUhFyGvo0KHuL7B//vmn+/G8eWNDhw4t9Pgbb7yx0ADd9RkeGhpa5BStyWRy/5s5cuSIz7+UXApkClL4jcPhcE/z7N+/3+sK31lZWZw7d67ALx5XYGC32zl9+jSxsbEcO3bM/S0v7xLyoipMezutVtQHct6RPFfCekl4e26Hw+H+88GDB91/L65id1lWnOUNDIsaSaxIxf0MR4wYwQcffICmaSxatCjf/dq6dSunT592t8srb/5bYQnNnsTGxnrdtjiHDh1y/9mbn2PHjh05dOgQqampxMXFeXxvFRXEldb111/PG2+8gaZpLF26NN9UvKvKe2Urs2AwGAgJCSEqKoo2bdrQu3dvhg0bhtVq9dk1Bg8ezO7du9m2bZt79N6lJKOCeT/PinsfmM1m2rdvz6ZNmzh06BCapmEwGDhy5Ii7TVE/h4CAAFq3bs3OnTsLPOf6N5GWllaiXRpiY2OLXZlZ3UgAJvzGVTG7NFJSUvL90sjKymLWrFksXLiQAwcO5AtMXAwGQ7HXM5vNXq9a9DRd6pJ3RZdrxKQkSnPuvKu1Chs9cynN5sIueUf3Kuu32OJWWdWpU4eePXvy119/sXjxYp555hl3AO+afgwICCiQh1NYHkxxLl4ZVhZ5+5C3REJh8o78pqSkeAzAymNVWu3atenSpQtbtmxh2bJl+QKwipx+BDyWMPGXIUOGMGXKFDRNY8WKFe60idjYWHbt2uVuU5y8X35K8j7QNI20tDTCw8PdnxkGg8HjSu+8CrtGZfg3camQAEz4Td4gqXfv3iVamZe38GdsbCz33Xcfx48fdz9mNptp1qwZzZs3p3379nTv3p3ffvut2OKflXUpvDfyTssWF/SV5XXmna6Kjo4ukKRenE8++YTQ0FB69OhBmzZtStwXX+2PN2LECHf9ti1bttC9e3fsdrt71dnAgQMLBCZ5RzN//vnnIqe88ypLwFtWeftcWG5jeb3vBw8ezJYtW9i1axenTp2iQYMG+aq8V8biq+XNNQ25e/duli1b5g7AXEFphw4diqzx51LSL3ae3gcXf5kr6n1Q2Hvd9Tl+2WWX8c4773jdn6q0w4O/SAAm/CbvKI2maaWeipg0aZI7+Bo6dCh33XUXHTp0KPCB8eOPP5a6r1VB3pyvvPWvPCntt1aAK6+8ksDAQLKystiwYUOxH9x5paen8/XXX5OTk4PVamXjxo0FcleK+8VSlmKYeV133XUEBweTkZHBokWL6N69O3/99Zd7ZOHi6Ucg3yhBnTp1KmQKJW8f8o56FiZvm+JGOXzNNQ2p6zrLly/n3nvvZfny5WiaRqtWrWjVqpVf+1NZuKYht2zZwvnz54mKinIH/t4GpXk/P0vyPjCbze5/c7Vq1QKcn79JSUlFflEoLN0gIiKCxMREcnJyKtV0clUkSfjCbywWi3ska/fu3R6nDfOaPXs2M2fO5I8//nC3/eeff9yrhnr06MGUKVPo3Lmzx29rrryeS1WbNm3cf96zZ0+RbXfv3l3q6wQFBbkTrE+dOsWqVau8Pvbnn392F9vt169fvuDLVRbEbrcXmjeXnZ1dpuAxr8DAQPfrWLVqVb7k+1q1anlMss8bMLjed4XZsWMH//d//8fixYs9Fggurbx5Np5yci7mKuoZFBTk94CxTp067sUzrrIZrkCjstb+8gfX+05VVVatWkVsbCy7d+9GURSv70tJ3gc2m839mdCkSRP3CJhrgQZQ5O4TmqYVug2T69/E8ePHi/3it2jRIr799ltWrlxJZmZmkW2rIwnAhF+5VhK6KpMX5vjx47z66qu8+uqrvPbaa+5f1nlXVxW1GfTJkyf5+++/3X8vTWJ8ZRcVFeWeHlyzZk2hI0W6rjN//vwyXev+++93/wxef/11r76Bx8bG5ttm58EHH8z3fN4aS3lrjeW1YcOGfFOtZeVKpE9MTGTLli3uGmrDhg3zWCgy78rX2bNnF3nud999lylTpjBx4kSfJuHXrFnT/ct3zZo1Rebhbdq0iaNHjwLOOnvelFfxNVewsX37dvbt2+f+d1gdpx9dGjduTLt27QDndkyuFZGdOnVyb/5dnHbt2rlHwRYsWFBkQLNgwQLS09MB3HtbgnOE0uWXX34p9Pg//vij0BEw178JXdeLnGVIS0vjxRdf5M033+Tpp5+u0uke5UUCMOFXd999t/sX3Ztvvsn+/fsLtMnKyuLpp5925/6MHTvW/VzeabeNGzd6HEVLTEzk8ccfz/eLu7KWTyircePGAc7yFC+++KLHQPPzzz8v0wgYOMuGuIprxsXFMXbs2AI1yfLau3cv48aNc/8SuPvuuwsEzHm/0XsqFHv27Fmfb2OUt8zEW2+95e7fxVsPuVx++eVceeWVgDMY/PLLLz22mzZtmns7q7Zt2/p8x4B7770XcL6PJ02a5LHUyenTp3n++ecBZ66P6xh/GzRoEIqioGkaL774ojvdoLA9KqsLV2C6ceNGfv/9d6Bko4Jms9n9WZiYmMjzzz/v8fNv37597tysgICAfLUS69evz6BBgwDnytR58+YVOD4+Pr7Irctuu+02d67kF198wcaNGwu00TSNyZMnu9+nt956a6m3urqUSQ6YKLWvvvrKq9VUrVq1cm9z0rRpUyZOnMj7779PcnIyt99+O3feeSd9+vTBYrFw4MABZsyYwbFjxwC44oor8n2AdO3alVq1apGYmOj+JT9mzBgaNGhAcnIyMTExzJ07t8C0la/yiCqbG264gXnz5rFu3TqWLFnC6dOnGTt2LE2aNCEhIYF58+YVmDIs7TfRhx9+mBMnTvDbb79x+PBhRo4cyYABAxgwYABNmzZFURROnjzJmjVrWLp0qfuXQ//+/T0Wdhw6dCiff/45DoeD7777joyMDG688UasVivbt2/n22+/JSEhgcaNGxeoK1VaiqIwYsQIPvvsM3dQ2rZt2yKX07/++uvccsstpKen88EHHxATE8Mtt9xCgwYNSEhIYMGCBe7pNrPZzKuvvurzb/s33XQTy5cvZ/Xq1cTExDB8+PACWxF999137lGLBx54wB04+ludOnXo3Lkzf//9t3uqrKyjX94me3fr1s29wX1lM3jwYP73v/9hs9nYv38/BoOhwJZnxXnggQdYs2YNu3btYsmSJRw5cqTAVkQzZ84kOzsbgOeff56mTZvmO8cLL7zA5s2bSU5O5vnnn2fz5s3uumHbt2/nm2++ITExkaCgII+jbGFhYbz22mtMnDgRm83G+PHjueWWWxgwYABhYWEcO3aM77//3v3vq2HDhkXuD1qdSQAmSu2nn37yqt2AAQPy7TN3//33oygKH3zwAdnZ2UybNs1dgTqvbt268emnn+bL77Jarbz33ns89NBDZGdns2XLlnwbabs0bdqUW2+9lffffx9w1lLKu7/fpeTDDz/koYceIiYmhh07dhTIVapZsyYjR47k66+/BijVZuHgXEn1zjvvcNlll/Hpp5+SmZnJihUrWLFihcf2ZrOZBx54gIcfftjjNlBNmjTh+eefd29h88svv+SbFjEYDDz55JMkJyd7fH+U1k033cRnn32W7+9Fadq0KT/88AP//ve/OXXqFH/99Rd//fVXgXbh4eG8//77Zdo/sDCKovDhhx/yn//8hwULFnDy5EmPoxQmk4knnniC8ePH+7wPJTF48OB8KQBlzf/y9ufvcDgqbQDWpEkT2rZt666j1bVr1xLn6FksFqZOncqTTz7J+vXr2b9/v3tf3byCgoJ46aWXPL63a9euzcyZM3nggQc4deoU8+fPL5CiMGrUKOLj491FVy92ww034HA4+O9//0tWVhZz5sxhzpw5Bdq1atWKL774grCwsBK9zupCAjBRIcaPH8+gQYOYOXMmGzZs4PTp02RlZREREUH79u0ZPnw4N9xwg8cclquvvppff/2VadOmsXHjRndOTGRkJK1atWLw4MEMHz4ch8PBZ599RlZWFkuWLMm3pc+lJCQkhO+++45ff/2V3377jf3795ORkUG9evUYOHCg+1uzS1E1x4qjKAr33XcfI0eOZOHChfz1118cPHiQpKQk7HY7oaGhtGjRgquvvppbb7212IKfY8aMoVOnTsyYMYPo6GjOnTtHREQE3bp1Y9y4cXTu3LlES9290bhxY7p27crWrVsxmUxFVgR3adu2LUuWLGHu3LmsWrWK/fv3k5KSgsVioWnTpvTr148xY8Z4VZ+ptKxWK++//z6jR4/mp59+YuvWrSQmJmK1WmnQoAG9e/fmtttu86qkQXkbNGgQb731Frqu06FDh3y7OVRngwcPdgdgpQ1KIyIimDZtGqtXr2b+/Pns2LGDc+fOERISQsOGDRkwYAC33HJLkRvCt2zZkgULFvDTTz+xaNEijh8/7l6pescddzBixIgCOZsXGz58OD179mTmzJmsW7eOEydOkJGRQUhICG3atGHIkCHcfPPNpf7CVx0oemmqRgohqpTp06e79wmcN29ekQsYhBBClD8ZAROiCps8eTKKonDllVfmm+a92KZNmwDntGDepehCCCEqhgRgQlRhycnJrFmzhiVLlnDVVVd5XNK+atUq9ya8ffr08ek+d0IIIUpHpiCFqMJWr17tLg9Ru3Ztxo4dS/v27QkJCSEhIYHVq1ezYMEC7HY7wcHB/Pbbb5KPI4QQlYAEYEJUcZ9++imfffZZkXsm1qtXjw8++IDOnTv7sWdCCCEKIwGYEJeAffv28eOPP7JlyxZOnTqF3W6nZs2aNGrUyL0qNG/leSGEEBVLAjAhhBBCCD+TrYiEEEIIIfxMVkFWIF3X0bSyD0AaDIpPziOKJ/faf+Re+4/ca/+Re+0/5XWvDQbFJ9uNSQBWgTRN5/z5gpvqloTJZCAyMpjU1EwcjsKTsEXZyb32H7nX/iP32n/kXvtPed7rqKhgjMayB2AyBSmEEEII4WcSgAkhhBBC+JkEYEIIIYQQfiYBmBBCCCGEn0kAJoQQQgjhZ5dkAPbwww9z2WWXsWXLljKdZ9OmTTz00ENcddVVdOjQgX79+vHcc89x4MABH/VUCCGEENXRJReAzZw5k1WrVpX5PDNmzGDcuHGsWbOGpKQk7HY7cXFxzJs3j1tuuYUFCxb4oLdCCCGEqI4uqQDs559/5rXXXivzeVavXs3bb78NwHXXXce8efPYuHEj06dPp02bNthsNp577jn27NlT5msJIYQQovq5JAIwm83Gq6++ygsvvEBZt7bUdZ3//e9/6LrONddcw8cff0z79u2JioqiZ8+ezJw5k2bNmmG325kyZYqPXoEQQgghqpMqH4CtWLGCoUOHMnPmTADat29fpvP99ddfHDx4EIDHHnsMgyH/LQoJCeHRRx8FYP369cTFxZXpekIIIYSofqr0VkSpqak88sgjAAQFBfHUU0/RsmVL7r777lKfc+3atQBERUXRsWNHj2369u2L0WhEVVVWr17NmDFjSn09X9M1jawD+3GkpGAKDyew9WUohsoVZ1eFPl4q5F4LIUTlVKUDMACj0cjQoUN54oknqF+/Pps3by7T+fbt2wdA27ZtC91sMyQkhEaNGnHs2DF27dpVpuv5UtrWLST+OBNHUpL7MVNkJLVGjyG0a7cK7NkFVaGPlwq51/4lwa4oiq5pZOzdj8ORRY4pEEuLVvL+KCdV5V5X6QAsMDCQ5cuX07BhQ5+d89SpUwA0aNCgyHb16tXj2LFjnDx50mfXLovULTHEffFpgccdSUnOxyc8UuG/dNO2bqn0fbxUyL32Lwl2RVHk/eE/VeleV+kAzGw2+zT4AkjK/aGFhYUV2S40NBRwToNWNF1VOZObA1eY+O+mo2ZmVti3AF3TOPvznCLbxH83A111oBiNoBicI5B5/lMMCijO/isGQ/7n8rU1ONuigCHvcwZQyP177vEGBSW3nftYRQGF/H0wKCgmIw6rAS0nB03VPVxXKXTU1J90TSPxx6LfD4k/ziKkc5dK+a2wqpFgVxRF3h/+U9XudZUOwMpDTk4OAAEBAUW2s1qt+dqXlslUtl+ARqOB1D17cSSdL7KdlpFBwrfTynSt8qZlpHPm/76s6G6UjSsAU5T8QSKuANJTwFjY313Hc1EQ6XrswvndAaeioGZn5/v254kj6TwnXn8ZY3Cw8zgu/A/yBJJ5HnO+LKXAY/n+nuf153/Ku+NcDykGAwlmI3aHinth80XBrfueeOhnvrZ5z31R/zwdp1x0nMfXl0sHUtb9SVHOTP+GnCOHCg/OS/J4UQF+Ec8VuHaevyuKQqrVRI7Nga7pHtsU/pji4SEvvoR4aFNUH0t0fi/Oo3h8/uLjCp63wJk89uXCY7qmkTh/XqFdBYj/dhpaWoqHL0PF30evvu/56kuh1+fx5ufvzeW8enHuP+maRvxPPxbZOnHOLCKu7FZpvnhKAHYRo9GIpml+uZbBoBAZGVzm8yQW88vWJbBRQ8wREWW+XmnYk5PJii1+utZaty7m0BDQdWdJEU1DB9C0C4/l/ud83sNjug665v5lousaaHmep6jj8v+5xFzH6Dr6Re+jshVI8b2cEycqugvVgp6dzfllSyu6G6KS0jIzif/h+4ruRrXgOH8e4+njhF/eoaK7AkgAVkBgYCB2u73YkS1vR8qKomk6qamZpT4enCNglshIr9rWvnMswW3blul6pZWxdy/H33mr2HZ1x92br4/56rp5CojyBDz5H/byuIv+fnEdOVcQiK5jQCc0JIC0tCxUu+PCc4CuqWi5QR6q6n6M3CDRHUiqap5ALzdIdAWXmnYhQHS1yX0OnYv+nhtk5g0Wc/uSExdH0qqVHu9vXhF9+mKpVSvPy3fdS/erz/O/gs8VqLnnuofk9lfx7lzu9nkeVFCwWk3k5NgvjMroF47T0fOf392Xgudyt/f2teV5XYW9RtefbYmJZB3YT3ECW7XCXLNW3hMVe4zn922xD3g8zuPVctspioLZbMRuUwv+O/K2X97w5jiPfS/pfVAKa+T5BXn1ejz0y4ufhSMpmZyTscWe3dqwIaa8X4599W3Ni9fm3aW87JBXzbx5H5T8PI7UVGynTxd7VNLJM2gNm3lzgUKFhQViNJZ9FE0CsIuEhoaSmppKWlpake1cuV+RXgY/hXE4yj7aFtGuLabIqCKnIU2RUVhatPLJ9UrD0qIVpsjIIqfGStdHL6YPfDQCbzAZsEQGY1AsaA6twGmNvrnMhV/6xQWOhQSdwZpK6tatqMlF3etIatw2utIMxV/MZDIQER5IcnJmOb5ny/5bLvPAAa8CsMghQwlq3drLs/rgDVuCaSeTyUBERJBv77Uvpr188u+2bCcpa05n5v59nPrg/WLb1bxtNEGXtfG2U2XqU6W5ho9l7t/HqSnvFttOCQmrsN+DF5MA7CLNmjXj1KlTxRZYPXPmDOBcDVnRFKORumPGcPLTTwptU2v0nRX6y1YxGKg1eozHBEmXiu5jZXEh/6pkH4IXWpupfUdx93oMxsDAUvXPHwwmA8bAQAzZGoZK8mHpSXCHy736YhHc4fJK+942mAwYAwIwWFUMxsp7r6uioLbtvHp/BLVtV2nfH1VF0GVtvLrXga0v82OviiY/8Yu0atUKgL179xbaJi0tjdhY57Byu3bt/NKv4oR1u5J6Ex7BdNGInCkyinqVZOVHaNdulb6Plwq51/7h+mJRFPliUX3J+8N/quK9lhGwi/Tt25fp06eTkJDArl276NChYLLe2rVrUXPzfHr16uXvLhYqtGs3Qjp3qdTFIKtCHy8Vcq/9I7RrN5jwiIfaQ1HUGn2nBLvVnLw//Keq3WsJwC7SvXt36tWrR1xcHO+++y7Tpk3DZLpwm9LT0/n0U+fUTr9+/WjatGkF9dQzxWAgqE3FJNp7qyr08VIh99o/JNgVRXG9P2yHD2Kt5NXZq7qqdK8rX4/8YOfOnQwePJjBgwfzww8/5HvOaDTy7LPPArB582YeeOABduzYQVJSEhs3buSuu+7i6NGjWCwW96bcQgjhCnbDelxFUJu2lfIDX1QcxWAguG1bavXpTXBbeX+Up6pyr6vlCFhWVhZHjx4FLlS+z2vIkCHs37+fL774gvXr17N+/fp8z5tMJt577z2P05NCCCGEEMWplgGYN5544gl69OjBd999x44dO0hJSSEyMpIrr7yS8ePH0759+4ruohBCCCGqKEUvUGlQ+Iuqapw/n1Gmc5hMBiIjg0lKyqg0tU0uVXKv/Ufutf/IvfYfudf+U573Oioq2CeFWCvnxKgQQgghxCVMAjAhhBBCCD+TAEwIIYQQws8kABNCCCGE8DMJwIQQQggh/EwCMCGEEEIIP5MATAghhBDCzyQAE0IIIYTwMwnAhBBCCCH8TAIwIYQQQgg/kwBMCCGEEMLPJAATQgghhPAzCcCEEEIIIfxMAjAhhBBCCD+TAEwIIYQQws8kABNCCCGE8DMJwIQQQggh/EwCMCGEEEIIP5MATAghhBDCzyQAE0IIIYTwMwnAhBBCCCH8TAIwIYQQQgg/kwBMCCGEEMLPJAATQgghhPAzCcCEEEIIIfxMAjAhhBBCCD+TAEwIIYQQws8kABNCCCGE8DMJwIQQQggh/EwCMCGEEEIIP5MATAghhBDCzyQAE0IIIYTwMwnAhBBCCCH8TAIwIYQQQgg/kwBMCCGEEMLPJAATQgghhPAzCcCEEEIIIfxMAjAhhBBCCD+TAEwIIYQQws8kABNCCCGE8DMJwIQQQggh/EwCMCGEEEIIP5MATAghhBDCzyQAE0IIIYTwMwnAhBBCCCH8TAIwIYQQQgg/kwBMCCGEEMLPJAATQgghhPAzCcCEEEIIIfxMAjAhhBBCCD+TAEwIIYQQws8kABNCCCGE8DMJwIQQQggh/EwCMCGEEEIIP5MATAghhBDCzyQAE0IIIYTwM1NFd6Csdu/ezdSpU4mOjiY5OZnIyEi6dOnCuHHj6NKlS6nPu3jxYubOncvu3bvJyMggIiKCTp06cdddd3HVVVf58BUIIYQQorpRdF3XK7oTpbV06VImTZqEw+Eo8JzBYGDSpEmMHz++ROdUVZWnn36aRYsWFdrmwQcf5MknnyxxfwteS+P8+YwyncNkMhAZGUxSUgYOh1bmPonCyb32H7nX/iP32n/kXvtPed7rqKhgjMayTyBW2SnIXbt28fTTT+NwOOjWrRuzZs1i48aNzJ49mx49eqBpGu+//z5//PFHic772WefuYOv/v3789NPP7F+/Xq+++47OnfuDMBXX33Fr7/+6uuXJIQQQohqosoGYB999BE2m41WrVoxbdo0unbtSlRUFF26dGHq1KlceeWV6LrOu+++i6Z5F/3abDZmzJgBQK9evfj888/p2LEjNWvWpEePHsyYMYOWLVsC8OWXX5bXSxNCCCHEJa5KBmCHDx/mzz//BGDChAlYrdZ8z5vNZp566il327///tur8x45coSMDOeU4C233IKiKPmeDwgIYMSIEQAcO3aMlJSUMr0OIYQQQlRPVTIAW7duHQBGo5E+ffp4bNOxY0dq1KgBwKpVq7w6r8Fw4Xaoquqxjcl0Yd1C3vZCCCGEEN6qkhHE3r17AWjUqBGhoaEe2yiKQps2bQDnSklvNGnShLCwMADmz59f4Hm73c7ChQsBaN26daHXFkIIIYQoSpUMwE6dOgVAgwYNimxXr149AE6ePOnVea1WK4888ggAf/31F4899hg7d+7k3LlzbN26lfvvv5/du3djNpt57rnnyvAKhBBCCFGdVck6YElJSQDu0arCuEaoUlNTvT73uHHjCAoK4qOPPmLZsmUsW7Ys3/NXXHEFzz33XJlqjAkhhBCiequSAVhOTg7gTIoviis539XeG6qqkp2dTVBQkMfnY2NjiYmJoWPHjhiNRq/PWxiTqWyDkK5aJL6oSSKKJvfaf+Re+4/ca/+Re+0/VeFeV8kAzBeBjycOh4MnnniCFStWYDabefzxx7npppuoVasWp06dYs6cOcyYMYP//e9/7Nu3jylTppQpEd9gUIiMDPZJ38PCAn1yHlE8udf+I/faf+Re+4/ca/+pzPe6SgZggYHOG1rcyJa3I2Uuv/76KytWrADgww8/ZODAge7nmjZtyrPPPkvLli15/vnnWbx4MQMGDGDo0KGleQkAaJpOampmqY8HZ3QfFhZIamoWqiqVlcuT3Gv/kXvtP3Kv/Ufutf+U570OCwv0ychalQzAXLldaWlpRbZz5X5FRkZ6dd6ffvoJgO7du+cLvvK65ZZbmDVrFrt27eLHH38sUwAG+GyLBFXVZGsLP5F77T9yr/1H7rX/yL32n8p8ryvv5GgRmjVrBkBcXFyR7c6cOQNcWA1ZnGPHjgG4txwqTPfu3QFn4VYhhBBCiJKqkgFYq1atADh+/DiZmZ6n8HRdd9cLa9eunVfntdvt+f5fHJvN5lU7IYQQQoi8qmQA1rdvX8AZKLmq4l9s+/btnD9/HoDevXt7dV7XyNqmTZuKbLd161YAmjdv7tV5hRBCCCHyqpIBWOPGjd3ThB9//LF7/0YXu93OlClTAGfF+quvvtqr8w4ZMgSAPXv28PPPP3ts8/vvv7Njxw6AMud/CSGEEKJ6qpIBGMBzzz2HoigcOnSIsWPHsmnTJpKSkti+fTvjx48nJiYGRVGYOHFivk21d+7cyeDBgxk8eDA//PBDvnOOHTuWFi1aAPDiiy/y6quvsnv3bpKTkzl48CDvvfcekydPBpzTmqNHj/bfCxZCCCHEJaNKroIE52bbr7zyCi+//DK7d+9m3LhxBdpMnjyZ/v3753ssKyuLo0ePAhcq6rsEBgby9ddfM2HCBPbv38/MmTOZOXNmgfNefvnlfP7551gsFh++IiGEEEJUF1U2AAMYNWoUHTp0YNq0aURHR5OUlERwcDCdO3dm3LhxXk895tWgQQPmzp3LvHnzWLx4Mfv37yc9PZ3Q0FDatGnD0KFDGTFiBGazuRxekRBCCCGqA0XXdb2iO1FdqarG+fMZxTcsgslkIDIymKSkjEpb6+RSIffaf+Re+4/ca/+Re+0/5Xmvo6KCfVKItcrmgAkhhBBCVFUSgAkhhBBC+JkEYEIIIYQQfiYBmBBCCCGEn0kAJoQQQgjhZxKACSGEEEL4mQRgQgghhBB+JgGYEEIIIYSfSQAmhBBCCOFnEoAJIYQQQviZBGBCCCGEEH4mAZgQQgghhJ9JACaEEEII4WcSgAkhhBBC+JkEYEIIIYQQfiYBmBBCCCGEn0kAJoQQQgjhZxKACSGEEEL4mQRgQgghhBB+JgGYEEIIIYSfSQAmhBBCCOFnEoAJIYQQQviZqaI7IIQQQghRFpquoes6qq6h6SqKDkqWgxyHihFzRXfPIwnAhBBCCFHp6bqOpmtoupYbaGk4NAcOzeEOvDRdQwdMJgMOsxWTbsWoSAAmhBBCCFGoi4MsHQ2HpuYGWqr7OQAdUBQdAwYMigGTwYRBMaAoCiajAWMlT7KSAEwIIYQQfpV3FEvTVVQt/2iWjoZzLAsUQMkNsowGI2bFjEFRKvYF+IAEYEIIIYTwOc0dYOWfMrRrDnfgpaGj6Ao6OgbFgEFRMBoMGBTnaNalTAIwIYQQQpTKxcnvmq5hzx3Jyht4Qe5IlqLkBloGjAYLxnIIsjRd43DyCTKS0wk3RNA28rJKGcxJACaEEEKIQrnystQ8I1qFJb9D/iDLbDBjQEHx05Th7nP7WHhkOam2NPdjEdZwbms1nE61L/dLH7wlAZgQQghRzfkq+b0i7T63j1n7finweHJOCl/v+p77O4ytVEGYBGBCCCFENXFx8rtDU1FzAy3X47pyIfndgAGlCiS/a7rGwiPLi2wz9+DvXFGrfaWZjpQATAghhLiElC753YDRYMBcRZPfj6acyDft6ElSTgqHko/SOrKFn3pVNAnAhBBCiCrmQpCllzj53WSwVMkgKy9VU4nLiOd42klOpMZyMPmoV8el5qSWc8+8JwGYEEIIUQlVpeT38pbtyOZE2imOp8ZyIu0ksWmnsWv2Ep8nzBpWDr0rHQnAhBBCiApSfPK7iqY7Q6zKmvzua7quk5yTwvHUkxxPi+VE6kniMxPcgaZLgDGAJmENaRLWkIYhDZh78PcipyEjreG0jGhWvp0vAQnAhBBCiHLmHL3SyHbkkGXPIttu9zL53YQ5t0DppUrVNc5kxHM81TmdeCwtljRbeoF2UQERNAltROOwhjQJa0StwJr57svQ5td7XAXpcmur4ZVq6lUCMCGEEMJHnEVJ1dwRLecqQ4fqwKGrKEadbKOV1OwsHKp2SSS/l0a2I4fYtFMcT4vleOpJTqadwnbRdKJBMVA/uC5NwhrSOLQRTcIaEmoJKfK87Wu04c42txSoAxZpDedWqQMmhBBCVH0X8rNU1Dy5Wa5pQ1XXUHCuMjTmycsKsgTiMIPDoFX0S/Ab93RiaizH02KJz0hEv2hCMcBopXFYQxqHOke3GobUx2I0l/ha7Wu0oW1Ua2LTT5KBVMIXQgghqqS8ifCukS27as+dOryQBK/kJrwbc3OzLB5ys4yXYL7WxVRdIz4jwT26dSI1lhQPeVmR1ojc0S1nwFU7qJbPplkNioEWEU0xBYGSY66UwRdIACaEEEIAztIGWt5AS3NgV+3uFYfOQAsU99Sh50CrOslx5BCbfto5upV6kti0U9g0W742BhTqhdSlSWhDGoc1okloQ8KsoRXU48pDAjAhhBDViqZrqJrqDqzsuUVKXQGYjp6vdlZlrwLvTyk5qe7VicdTYzmTkVBgOtFqtNI4tIEzWT60EQ1D62M1Wiqox5WXBGBCCCEuSe6pQ801omXPVw1e152lHQwKeZLhJdBy0XSN+MxE9+jWibRYkj0UMo2whueObjmnE+sE1arwaT9N19F154hlZSUBmBBCiCotb7FSV2kHu3Zh6lDD+cvY4BrRukSqwfuaTbURm5Y7nZjmnE7MUXPytVFQqBdcJ9/qxPBKUtzUVahW1VSMJgMhxjA0xVjR3SqUBGBCCCGqhPz7G6o4VDU30MqdTuTC1jsGxYhBMWCRQKtQKTmpnEg76V6heCYjHu2i6USLwULjsAbuZPlGIfWxmqwV1OOCLqxAVTGgYDaYCLQEEmwNoFZIOCmOLByOyrniVAIwIYQQlUr+Wlr5Szy4HgdQFGfBUoPBiNlgxiiBVqE0XSMhMzE3f8u5OjEpJ6VAu3BLmHN0Kzd/q25w7UoXwLp2CtB0DYOiYDaYCbUGYzKYMRmcgbfJaKh0/b6YzwMwVVX5559/iI2NJTU1lRtuuIHIyEiSk5PJysqiXr16vr6kEEKIKshTiQebaiu6lpbx0trjsLzYVBsn005zPM05uhWbdopsD9OJdYNru0e3moQ1JMIaXkE9LpwrIHcGXc73gtVgwWqyYjaYMBlMVfL94LMALDs7m88//5zZs2eTnn5hC4GuXbsSGRlJdHQ0jz/+OAMHDuS5556jfv36vrq0EEKISiz/ptLOwKq0tbSEZ6m2NE6knnSvUIzLiHePFLpYDGYa5Vmd2Ci0AQGVaDoxL13XcegqDs0OujN3z2q0EmAKcAddVZ1PXkFiYiL33XcfBw8eRNcvzB/n/Ydz6tQpdF1n5cqVxMTEMH36dNq2beuLywshhKgkvK+lpWBQjFJLqxQ0XScx62y+1Ynns5MLtAuzhF6ovRXWkLrBdSr1NK2m6+4kel3RMSkmgkxBWI0W5xSzofIm1JdGmQMwXdd57LHHOHDgAADdu3enb9++vPfee/natWzZkoYNG3Ly5EmSk5P597//zcKFCwkKCiprF4QQQvhZcbW0AHT0CysPq3ktLU3XOJx8AjXdhtFhoVFIQ69zlOyqnZPpp92jWydST5GtZudrowB1gmrTOKwhTcMa0Ti0ERHWsEof2OZduQhgMpgJNgdhMVowG0yXXNCVV5kDsCVLlrBt2zYUReHFF1/kzjvvBCgQgPXu3Zvly5fz5ptv8sMPPxAXF8ecOXO49957y9oFIYQQ5aS4Wlqanjt5qDi32pFaWgXtPrevwAbRYZZQhja/nvY12hRon25Lz83dcibLn844g3rRdKLZYKZRaH0ahzoDLud0YkC5vxZfUHUNh2rHoWv5Vi5ajRZMhuqzKXmZA7AFCxYAMHDgQHfwVRiDwcALL7zAwYMH2bx5MytXrpQATAghKgHXdKErsLKrDneOliq1tEpt97l9zNr3S4HHU21pzNr3C6Mvu5k6QTXzrU48l51UoH2oOYQmuYVOG4c2pF5wnSo1OuRx5aIpAJPBjLmKJtGXVZkDsF27dqEoCkOHDvX6mFGjRrF582YOHTpU1ssLIYQogYtraaHqqJk2zmelkmN3SC0tH9J0jYVHlhfZ5sf98wo8pgC1g2o5Vybm5nBFWsOrVJByqa5c9KUyB2DJyckAJVrV2LBhQwAyMjLKenkhhBAeeFtLy2Q0YFJDQAezUWpp+Yqu6+w5dyDftGNhjIrBXQqicVhDGoc2JLCKTCfmVR1WLvpSme9GSEgIycnJ+UpPFCcxMRGAsLDKsX2BEEJUVSWppeWaOjTn5tkoioLJaCDAZCXHqOJQK2fF8MrOrtqJz0wkPjORMxkJnMlMID4zgQx7plfHj2w5lM61Ly/nXpaP6rZy0ZfKHIA1b96cv//+m02bNnH11Vd7dcyiRYsAaNGiRVkvL4QQ1YL3tbRAyS1aKrW0fEvTdZJzUjiTEe8MsjISOZOZwLms8+gXbeFTEpVlL0VvFbVy0WI0y3S1l8ocgA0YMICtW7cyc+ZMbrvtNvf0YmEWLlzIkiVLUBSFfv36lfXyQghxySm6lpaGjl6glpZZMcjKQx/KcmQTnzuadSbDOaIVn5lIjmrz2D7IFEjd4DrUDapFneDa1A2qTc3AGny07asipyHDLWE0DWtUXi/DZ2Tlou+VOQC78847mTFjBomJiYwdO5bJkyfTs2dP9/Oub1579+5l1qxZ/PLLL+i6TmRkJKNHjy7r5dm9ezdTp04lOjqa5ORkIiMj6dKlC+PGjaNLly6lPm9CQgLTp09n7dq1nD59GkVRaNq0KUOGDOHuu+8mIKDqzc8LISqXwmppuR4HqaVV3lRd41zWOffUoSvYSs5J9djeqBipHVSTOkG1qZsbaNUJrkWoOcTjSOPQ5td7XAXpcmPz6ypt8CIrF8uXouctXV9KO3fuZNy4cWRlZRX4gURGRpKZmUlOjnMPKl3XMZlMfP31115PWRZm6dKlTJo0CYfDUeA5g8HApEmTGD9+fInPu3btWiZOnFjoIoHWrVvz7bffEhUVVeJz56WqGufPl20hgslkIDIymKSkjEq74/ulQu61/1xq9/riWloOzYFdu1Ad/uJaWoY8/5U3k9FAeEQQKcmZl3QOmK7rpNszLuRo5f4/IfMsqq56PCbcEuYMsoJruwOumgFRJc5r8lQHLNwSxo3Nr/NYB6yieFq5aKmiKxfL8zMkKioYo7Hs/zZ9EoABHDx4kKeffpp9+/ZdOLmicPHpGzVqxDvvvFOm0Slwlr+44447sNlsdOvWjSeffJJmzZpx7NgxPvzwQzZv3oyiKHz55Zclmurcu3cvt99+OzabjQYNGvDUU0/RrVs3zp8/zy+//ML333+Pruv069ePr776qkyvQQKwqkXutf9U1XvtWgXmqqXlUNXcQMtzLS1XUnxFjoBcigFYSZPiLQYLdYNrFRjVCjQF+qxPmq4Rm34S1VTySvjlSc9NonfoDvfKRYvRXOVXLlarAMxlw4YN/PHHH+zdu5ekpCQcDgcRERG0atWKXr16MXDgQIzGsq+KuP/++/nzzz9p1aoVv/zyC1brhQ1F7XY79957LzExMbRo0YKFCxdiMHh3s26//XZ27NhBvXr1mD17NvXq1cv3/IcffsgXX3wBOIvQtm7dutSvQQKwqkXutf9UtXvtXHloJ8ueRY5my19LCwMGg9EdbFU2VTkAK2lSvALUCIhyjmjlBlp1g2sTYY3wy7RuZbnXF69cNCsmLEbLJbVysSoEYGUObQ8cOECjRo0IDHR+U+jZs2e+HLDycPjwYf78808AJkyYkC/4AjCbzTz11FOMGjWKw4cP8/fff9OtW7diz7tz50527NgBwHPPPVcg+AIYO3YsX331FRaLhZ07d5YpABNCVG2arpGj5pBhz8Km2VBQsBjM7hIPwnc8JcWfyUjEpnmfFF87qBYWo9nPPa8cNHctOBUF58rFEHMwZqNZVi5WkDIHYC+//DL79+/n0Ucf5Z577vFBl4q3bt06AIxGI3369PHYpmPHjtSoUYNz586xatUqrwKwJUuWANC4cWMGDRrksU2NGjXYtm2bJOELUY2pmuoOvOyaHaNiIMAYIMnxPlDeSfHViaeVi2GWICxGs6xcrATKHIAdPnyYzMzMMiekl8TevXsBZz5ZaGioxzaKotCmTRvWr1/P7t27vTrvzp07AejRo0e+x3VdR9M099SpBF9CVE921U62I4csNRuH5sBkMBJoCqj2v+hLoyKT4i9l+VcuOovuysrFyqnMAVh2djbgHDXyl1OnTgHQoEGDItu5phBPnjzp1XkPHjwIQJMmTdB1nblz5/LTTz+xb98+HA4HjRs3Zvjw4fzrX/9yT7kKIS5tuq5j1+xkObLJcmSj6Tpmo4kgs3wGeKsyJsVfKorcc9FoxqQYJeiqpMocgLVt25YdO3awZcsWOnXq5IMuFS8pyblTfHFbGblGx1JTPQ9d52Wz2UhJSQEgKCiI8ePH89dff+Vrc+zYMT7++GOWLVvGN998Q+3atUvTfSFEFZA3sT5bdZbRMRvNmGS0pVBVLSm+qsq7clHXndOwsudi1VPmn9Lzzz/P3XffzaeffkpERAQ33XQTJlP5/vBdNcWKmwp0Jee72hcl716WX3zxBYmJidxwww089NBDNGvWjISEBGbPns3UqVPZv38/jz32GLNnzy7zNwuTqWxz8K6VGL5YkSGKJvfafyryXmu6Ro4jhwxHFjaHDUVRCLIGVMoVjL5Q2nud5ch2jmZlJBCXEU9c7p9tRVSKrxdSh7rBtakXXId6wdUvKb4s72v3ykVdRdd1zCYTQcaQS2rloi9Vhc/rMkdKiYmJ3HfffXzxxRe8+OKLvPHGG7Rr1466desSElJ0EqSiKLz00kslvqYvylhcLG+QlpiYyKhRo3j11VfdjzVs2JCnn36ayMhI3nvvPbZt28aKFSu4/vrrS31Ng0EhMjK4TP12CQuToXl/kXvtP/681w5NJduRTUZONqrBTnCghUhjcLVJVA4J9fyFVtVU4jPOcir1TO5/cZxKjed8VrLH9iaDkXohtakfVpcGYXVpmPv/MGuoTIXlKuxeX0zTNeyqA4fqAEUhxBBEgDkAq9GMxWjxurxSdVaZP6/LHID9+9//zvePKisri7///tvr40sTgLnyr4ob2fJ2pOziNhaLhaeeespju3vuuYfp06dz9uxZli1bVqYATNN0UlM950B4y2g0EBYWSGpqFmoVq+FT1ci99h9/3mt3Yr0jC7vmwGwwYTaYcSgqOXhOBr+UGI0GQkIDSEvNIiU7jbj0C6NZZzLiiS8qKd4a5h7NciXH1wqsUXA0JhtSs7P88GoqN9e9Tk/LLvR9reYm0Tt01Z1EH2AKcK9c1BwGslDJQu5nUcrzMyQsLLBy1AEDClS797a2a2m/Dblyu9LSCt/gFC7kfkVGRhZ7zuDgCyNRbdu2LTS/zGQy0alTJ1auXMnhw4e97XKhfFUgTlW1KlGw8lIg99p/yutee0ysNxixGp1fxFRNBw85S5eKvEnxCVmJJOac5WRKnE+S4nWdKlfQ1d9UVct3jzyvXAzKv3JRB03FXeRXeKcyf16XOQBbtWqVL/pRIs2aNSM6Opq4uLgi2505cwbAY0HVi1ksFmrXrk1CQkKBwq4XCwkJAbzLLRNCVB7VLbFekuIrJ1cSvV1zoMvKxWqrzAFYcaUgykOrVq0AOH78OJmZmQQFBRVoo+u6u15Yu3btvDpvmzZtSEhIIDY2tsh2586dA5BVkEJUEXkr1ts1OwAWk+WSSqwvbaX4eiG1aV6rERFKBFHWGtUqKd5f3Jux6zoZNufPStcUAk0BWI1WWblYTVXJn3jfvn15/fXXsdvtrFu3zmPV+u3bt3P+/HkAevfu7fV5//zzT+Li4ti5cydXXHFFgTaZmZls27YNwG9lN4QQpeOpYr3VaK3Sozm+rhRfWfYnvBTouu4OtjRdQ9c1QEFRwKAYMZnMhFtDMQVaUTSjrFys5nwegB0+fJh169Zx4MABUlJSMBgMhIeH07x5c3r27EmbNm3KfI3GjRvTuXNntm3bxscff0yvXr3y5XDZ7XamTJkCQOvWrbn66qu9Ou8NN9zA//73PzIyMnj99df57rvvCiTwf/LJJ6Snp6MoCsOHDy/zaxFC+J5dtZOt5pDlyMau2TEbTOVesV7TNY6lxpJmSyfUEkLTsEZlWkEpleIrNy032NJ0NTfg0p2BFobculwWLEbnKKtRcQZbFrOJsIBg1Kyqscm8KF8+C8DOnj3Lf//7X9asWVNku549e/LWW2+VefruueeeY9SoURw6dIixY8fyzDPPcNlll3H8+HE++OADYmJiUBSFiRMn5vvQ3blzJ8888wwAd911F3fddZf7uaioKJ555hleeuklduzYwZ133snjjz/O5Zdfzvnz55kxYwY///wzAHfffTctWrQo02sQQvhOYYn1weaCKQq+tvvcPhYeWU6q7cLCoDBLKEObX0/7GsV/6fRUKf5MRgKZDqkUXxm4R7U0Z7AFznw5Q25g5ZxGdNbicgVcksMliuOTACw2NpYxY8aQmJhY7ArIDRs2MHLkSH788UcaNWpU6mt27NiRV155hZdffpndu3czbty4Am0mT55M//798z2WlZXF0aNHgQsV9fMaPXo0qampfPjhh+zevZsHHnigQJsbb7yRSZMmlbrvQgjfcSfWO7LIdvg/sX73uX3M2vdLgcdTbWnM2vcLd7a5xR2EFUyKT+BMZqIkxVcS+acQVeeoVu4UolExYjaYCcotB2FUcoMtGV0UpVTmAExVVR5++GESEhIA6N69O3feeSedOnWiRo0aaJrG2bNn2bFjB7NnzyYmJoZz584xceJE5syZU6aiqqNGjaJDhw5MmzaN6OhokpKSCA4OpnPnzowbN87rqceLPfDAA/Tp04dvv/2WzZs3k5iYSFhYGG3atOH222/3mHMmhPCvypBYr+kaC48sL7LNvIOLOHD+MAlZiV4lxdcNquUOtqpbpXh/ck4hqhfytdBBB6NiwKAYCTA6a28Z8gRa1aUor/APRfe2aFch5s2bx/PPP4+iKDz66KM8/PDDRbb/7LPP+OSTT1AUhY8++qhMhUyrOlXVOH8+o0znMJkMREYGk5QkOQXlTe61/xR1rz0l1puNlgoZETqScpypu34o0THFJcX7W3VIwncFWYVNIVoMuaNa5TyFKJ8h/lOe9zoqKrhyFGJdunQpANdcc02xwRc4K+f//fffrF+/nl9//bVaB2BCCO/ZNQfZufld/kqsL05qTtHFoF3aRrXi8prtJSm+nHmaQgQwKApGxYjFYMFkNMkUoqgUyhyA7dmzB0VRuOWWW7w+5rbbbmP9+vUcPHiwrJcXQlzCdF3HptrcifWqrmExmPySWF8Uh6ay6+xe1sSu86p9z/o9aB7epJx7Vb3kn0JUcc3luEawrKZAzAaTe5TLObUoU4ii8ihzAJacnAw4N6v2Vv369QHceWNCCJGXpmtk27NJzk4hPce5553ZaCaggkcrMuwZRJ/Zxua4raTZ0706JtwSRtOw0i84EuRbgajrOjrO5HhD7ghWoDEAs9E5qmWQVYiiiihzABYcHExqaqq7Orw3XAVSPVWwF0JUX67E+hx7DllGE9lqTqWoWH8mI4ENp6PZkbgLR24NrlBzCD3qdSXMEsq8QwsLPfbG5tfJyIuXdF2/kK+lq2g6oOgYUNy1tS6Ue5ApRFG1lTkAa926NVu2bGHp0qX069fPq2NceWNSR0sIAQUT660mEwHmYOwmvcISwzVdZ3/SQTaejuFwyjH34w1C6nFN/e60r9HWXeoiwGQtUAcs3BLGjc2v86oOWHUkU4iiuitzANa/f39iYmL4/fffGTBgANddd12R7VeuXMnvv/+Ooihce+21Zb28EKIKKyyx3myquCX/OY4c/k7Yyca4GM5lO2sFKii0r9GGnvW70zi0QYHprfY12tA2qrVPK+FfSvJPIWroOO+p0WDEpBgx55lCdE0jyhSiuNSVOQC77bbbmDZtGmfPnuWJJ57gjjvuYNSoUe4Ns10OHTrEjz/+yOzZs9E0jRo1anDHHXeU9fJCiCrm4or1lSWx/nx2MpviYtgSv4Mc1VnQNcAYwJV1O3NVva5EWMOLPN6gGKp9ov2FKURXsEXuFKLn7XkMMoUoqrEyB2AhISG89957PPTQQ2RnZzNz5kxmzpxJUFAQkZGRgLPifGamc0sNXdcxm8188MEH+fZvFEJc2gqrWF+RifW6rnMsNZYNp6PZe/6Auxp9zcAa9Kx3JZ1rX47FaKmw/lVmeTed1nQVdAUdPc8UonN7HleQJVOIQuTnk62IrrrqKqZPn87zzz/v3uYnIyODjIyCRUabNGnCu+++S8eOHX1xaSFEJVcZKtZfzKE52Hl2DxtORxOXEe9+vFVEc66ufyWtIlrIVj95FDWFaFZMmI2BmIxGmUIUogR8thl3586dWbRoEatXr+avv/7i4MGDJCcno+s64eHhXHbZZfTq1Yv+/ftjMMi3ICEudZ4q1luN1goNbNJt6Ww+8zfRZ/4m3e78gmg2mOhU63J61r+S2kG1KqxvlcHFU4iarqMoFDqFKNvzCFF6PgvAAAwGAwMHDmTgwIG+PK0QogqpjBXrT6efYWNcDDsSd6PmlpEIs4RyVb1uXFmnE0EVnH9WEVxTiKquk2GDDHsmmqq7Vx1ajdY8JR+c04cSbAnhOz4NwE6fPs369eu57bbbCjz36aefomkaI0eOpFEjKUooxKWkMibWa7rGvvMH2XA6hqOpx92PNwptQM963Wlf47JqkQCef3se5xQiKM6RLcWIyWQm3BqKMdCKriruYEumEIUoXz4JwOx2O2+//TZz5sxB13WGDx+O1WrN12bVqlXs27eP//u//+PBBx/k0Ucf9cWlhRAVqDIm1mc7ctiasIONp2NIykkGwIBCh5ptuTq3jMSlSssNttyrEHMXFbimEAOMVsxGc74pRIvZRFhAMGpWBg5kg2gh/MUnAdikSZNYsWIFuq6jKApHjx6lTZv8xQfT0tLQdR2Hw8Hnn39Oeno6zz33nC8uL4TwM2divY0Me2alSaw/l5XEprgYtibsIEe1ARBoCuTKOs4yEuHWsArrW3lwj2rlJscDKFDoFKJszyNE5VLmAGzlypUsX74cRVFo1aoVEydOLFADDGDFihXs2LGDV199lT179vDdd99x/fXX07Vr17J2QQjhJ5UtsV7XdY6mHGd9XDT7zx/MHe+BWoE16Vn/SjrVuhyL0VwhffOV/FOIqjMxPncK0agYsRgsmIwmTAaTbM8jRBVS5gBs7ty5ADRr1ow5c+YUur+joih06tSJH3/8kWHDhnHixAlmzZolAZgQVUBlS6y3aw52Ju5mw+lozmQmuB9vHdmCnvW60zKiWZUc7SmwPQ+ATm5elpEAYwAWo9k5ypUbaElivBBVU5kDsF27dqEoCvfff79Xm2tbLBbuu+8+/vvf/7J169ayXl4IUU4qY2J9mi2dzWe2En3mbzLszuLOZoOZLrWv4Op63agVVLPC+lZS7iKmhUwhBuZuz+MOtmQKUYhLSpkDsOTkZKBkG2u3bNkSgHPnzpX18kIIH6uMifWn0uPYcDqGf87udgcr4ZYwrq7XjW51OxFoCqywvhXH0xQigEFRZApRiGqszAFYVFQUiYmJnD9/3utjsrKyALwaMRNC+IcrsT7TnomtEiTWa7rGrrN7WXdyM8dSY92PNw5tSM/63WlX47IKTfr3xFnINM8UYm5S2oXteQIxG0zuUS7ZnkeI6qvMAViTJk1ITExk2bJl9OvXz6tjVq1aBUDTpk3LenkhRBldnFhvUAxYjZYKCwyyHNlsi9vBpi1bOZeZBDg3ur68Zjt61ruShqH1K6RfRXGNGmq65i7v4JpCdG86LVOIQog8yhyA3XjjjcTExPD777/Tv39/rrvuuiLbb9iwgZ9++glFURgwYEBZLy+EKKXKllh/Nus8G+Ni+Dt+h3sELsgUSPe6XehRtyth1tAK6VdRHJoDu+oAwGq0EGgOxJJb+kEIIYpS5gBs2LBhfPnll8THx/P4448zYsQIRowYQfv27QkNdX5gpqWlsW/fPhYsWMC8efNwOBxERUVx5513lvkFCCG8V9kS63Vd53DKMTacjuZA0iF3GYm6QbUY2Ko3bUJao1C5ghkt9x6qmopRMRJkDiTQFIDZYJYRLiGE18ocgAUHB/PRRx9x9913k5OTw/z585k/fz6Ae9NtTbtQXVnXdaxWKx9//DEhISFlvbwQwguVLbHertrZnriLDXExJGQmuh+/LLIl19TvTuuo5kREBpOSnIlDrRzV2VVNxabZ0dGxKGZCrcFYjBZMBp/u6CaEqCZ88snRsWNH5s2bxwsvvMC2bdvcj6uqWqBthw4deP311wtUyhdClI9sRzYZlSSxPiUnlc1nthJzZhuZDudiHIvBTJc6Hbm6XjdqBtYAqDQjSbquO6cZNRWDAgFGK4GmwNxaXJI8L4QoPZ99dWvRogWzZ8/m8OHDrF69mmPHjnHu3DkcDgfh4eG0bNmSnj170rFjR19dUghRBFVTybBnkmHPQKngxPqTaafZcDqaf87tRcstIxFhDefqelfStU5HAk0BFdKvwmi6hl21o+oaZoOJUEuwex9FIYTwBZ+Pnbdo0aJENcGEEL6Xo9pIs6VjU22502T+n2pUdY095/az4XQ0J9JOuh9vGtaInvW70yaqdaUrI3EhqV7HYrQQZg7EYrBIUr0QwufKNXkhPj6emJgY4uPjqVOnDl27dqVevXrleUkhqjVN19yjXkCFrGrMcmQRc2Y7m+K2kGJLBZx1sK6o2Z6r619Jg5DK9RngWpjg0FSMioEgcyABpgAsklQvhChHpQrA4uPjmTlzJvv372fixIkF8rl0XefNN9/kxx9/xOFwuB83Go0MGzaMF154geDg4LL1XAiRj021k2HLIEvNxmI0+z05PDHzLBviYtiW8A/23HyzYHOQu4xEqKVyLbpRdQ27akPTdcwGMxHWYMxGC2ZJqhdC+EGJP2lmzZrFW2+95Q6sbr/99gIB2KRJk1iyZAm6qwx0LofDwfz589m3bx/Tp08nIiKi9D0XQgDOUa9sRzZptnR0dAJMARj8NHKj6zqHko+yIS6aA0mH3Y/XDapNz/rduaJW+0oV0LiS6h2aA0VRsBqsBJoDsFRgfpwQonoq0SfjL7/8wquvvoqiKOi6jslkIicnJ1+bpUuXsnjxYsC5kqlHjx6MHTuWoKAg/vjjD2bOnMm+fft48803effdd333SoSohuyag3RbOtmOHEwGo9+SxG2qne2J/7DhdAyJWWcB50bSbaJa07N+d5qFNa5U03earmNXbai6hslgIsQcgtVkxWwwVap+CiGqD68DsLS0NN5//30AQkNDmTRpEsOHDycwMP8muP/73/+AC8HX1KlTMRqdCaxXX301bdu2ZfLkySxYsIBx48bRvn17X70WIaoNXdfJVrNJs2Wgag6sJqtfRnBSclLZFLeVmPhtZOWWkbAaLXSt3ZGr6l1JjcDIcu9DSTg0FbtmAxTMBjOhpkCsRkmqF0JUPK8DsCVLlpCUlITZbGbatGl06NChQJudO3dy4sQJ99+feeYZd/DlctNNN/Hjjz+yY8cOFi9eLAGYECWkairp9gwyHJmYFCOB5sDiDyqjE2mn2Hg6ml1n96Ll1quPCojg6npX0qV2RwJM1nLvg7ecSfUO7JrDmVRvCiLAFJC7CbZMMwohKgevA7A///wTRVEYNmyYx+AL4I8//gCco18tW7akXbt2HtsNHjyY7du3s3HjxpL3WIhqStd1ctQc0mwZ2DQ7ASZruZZxUDWV3ef2seF0NLHpp92PNwtrQs/6V9ImqlWlCmjU3Npduq5jMpgIt4S6pxmFEKKy8fqT6eDBgwD07t270DZ5A6pevXoV2q5169YAJCQkeHt5Iao116hXpj0Tg2IgqBzLS2TaM4mJd5aRSLWlAWBUjHSs1Z6r611J/ZC65XLd0rJrDhyqHVAIMFoJMAdUaNFZIYTwhtcB2Llz5wCoX7++x+dtNhu7du1y//2qq64q9FyuTbpTUlK8vbwQ1Za/iqomZCay4XQM2xP/wa45VzmHmIPpUbcr3et2JqQSlZHQ3LW7HJgUE0HmINkQWwhRpXgdgNntzro+F+d0uWzbti1fm65duxZ6rtRUZ3FG2YxbiML5oqiqpmscS40lzZZOqCWEpmGN8o0MabrOweTDbDgdw6HkI+7H6wXX4Zr63bm8ZrtKtdm0a0NsALNiItQaIUn1QogqyetP1ho1ahAXF8f58+c9Pr9p0ybAmf/Vrl27IoOro0ePAhAZWblWTAlRWfiiqOruc/tYeGS5exoRIMwSytDm19MqojnbEv5hQ1wMZ7Oco9sKCu1qtObqet1pGtao0owkXZxULxtiCyEuBV5/qjdp0oS4uDj++ecf+vTpU+D5lStXuv9cVJ4YwKpVq1AUhebNm5egq0Jc+vIWVdXQSl1Udfe5fcza90uBx1Ntacza9wtmg9ldrd5qtNKtTieuqteNqICIsr4En3HeixxsDgdmg4kwS4hsiC2EuGR4HYD16dOHjRs38ssvv3DfffcREBDgfm7r1q3uJH2A66+/vtDzbNmyhY0bN6IoSpGJ+kJUN66iqlmObMwGE1Zj6Uo7aLrGwiPLi7mWnUhrBNfU706X2ldgrURlJByaA7uuYbKD2Wgm2BQsG2ILIS45Xo/fDx06lKCgIOLi4njggQc4duwYmqaxZcsWnn32WcA5/di5c2cuu+wyj+c4ceIEzzzzDAABAQEMHDjQBy9BiKpN13WyHFkkZSeT7cgmwFS2UZ5jqbH5ph0LM7LljVxd/8pKEXxpuo5NtZFpz0LVVILMgdQOrklUQASBpkAJvoQQlxyvR8Bq1arFY489xttvv01MTAxDhgxxb0nkYrFYeO211/Idl52dzZYtW/jzzz+ZO3cumZmZKIrC+PHjqVmzpu9eiRBVkENzOBPtfVhUNc2W7lW79Nzk/orkSqrX0bEozg2xLUYLARYLVpOFTMUO6MWeRwghqpoSZfbec8892O12PvroIxwOR77gKygoiA8//JAWLVrkO+bgwYPcf//9AO72AwcO5MEHHyxr34WossqrqKqqa5xIjfWqbWgFlZVwbYht1xwYFCU3qV42xBZCVC8lXlp1//33c8MNNzB37lwOHToEQNu2bbn99ts9jmjVqFHDHXiZTCbGjRvHpEmTMBjkg1ZUT+VVVPVE6kl+O7KEMxnFFzgOt4TRNKxRma9ZElpupXrXhtihsiG2EKIaK1WBnwYNGvD444971bZmzZo8+OCDNGnShH79+hEVFVWaSwpxSSiPoqoZ9gyWHVvD1oQdgLNeWIcabYmJ31boMTc2v85vo03uDbF1MBsthJkDJaleCFHtlXuFRYvFwsSJE8v7MkJUar4oqlrwnDpb4rex/PgashzZAHSpfQWDm/Yn2BxMq8jmBeqAhVvCuLH5dbSv0aZM1y6Onlup3q6qmAwXNsS2SKV6IYQA/BCACVHd2VQ76bZ0stWcUhdVvdip9Dh+O7yEU+lxANQNqs3wFoNpkmdasX2NNrSNal1kJXxfc26IbUPTdcwGM+HWINkQWwghPJBPRSHKia+KquaV5chixfG1RJ/Zig5YjRYGNu5Lj3rdPCbxGxQDzcOblOma3pANsYUQomQkABOiHPiqqKqLrutsS/yHpcdWkWHPBKBjzfYMbjaAMEuoL7pcYq4NsVVNxWQwEWwOJsAUIEn1QgjhBQnAhPAhXdfJVrNJs2Wgag4CTNYyjwKdyUjg9yNLOZ5bXqJWYE2GNx9E84imPuhxybmT6lEwG8yEWkNkQ2whhCghCcCE8BFfF1XNduSwOvZPNp6OQcOZU9W/US961u/hk9WTJXHxhtiBpkACjAGyIbYQQpSSBGBClFHeoqp2zYG1jEVVdV3nn7N7WHx0JWl2Z1X79jUu44Zm1xFhDfdVt72i6Ro21Y6ma5gNJsItoViNFtkQWwghykgCMCHKoEBRVXNA8QcVITHzLL8fWcaRlGMARAVEMqz5IFpHtij6QB9zaA7sqgNwJvoHmgOxGMwyzSiEED4iAZgQpZS3qGpZc6Bsqo01setZf3qTu1J83wY96d3war+VcMibVG9UjASZAwk0BWCW2l1CCOFzEoAJUUKarpFmS/dJUVVd19l7/gCLji4nOScVgMsiWzK0+fVEBUT6rM9FuXhD7NDcDbF9Ua9MCCGEZ/IJK0QJ2Bw2krJTyLBllbmo6rmsJBYeXcaBpMMARFjDubHZdbSNal3uI04XNsRWMSjIhthCCOFnEoAJ4QVN10i3ZZGVqZPjyClTUVW75uDPkxv48+QGHLqKUTHQq8FV9GvYC0s5J7fn3RDbbDARagkmwGiVpHohhPCzKh+A7d69m6lTpxIdHU1ycjKRkZF06dKFcePG0aVLF59d5+DBg9x8883YbDZmzpxJt27dfHZuUbnZVTvptgxs5FAzOJwgcyAOVSvVufYnHWLhkeWcz04CoEV4U4Y1H0StoJq+7HIBF5LqdSyyIbYQQlS4Kh2ALV26lEmTJuFwONyPJSQksHTpUpYvX86kSZMYP358ma9jt9t5+umnsdlsZT6XqDouLqoabA3EYjSThb3E50rOSWHRkRXsOb8fgFBLCDc0vY7La7Yt1+lGh+YgR7VjUgwEmQNlQ2whhKgkqmwAtmvXLp5++mkcDgfdunXjySefpFmzZhw7dowPP/yQzZs38/7779OyZUv69etXpmt9+umn7N271zcdF1WCp6KqpcmNcmgq609vZk3sX9g1OwYUrq7fnQGNemM1lW17ouKua1NtGBUDYeYQAswBsiG2EEJUIlX2E/mjjz7CZrPRqlUrpk2bhtXq/GUWFRXF1KlTuffee4mJieHdd9+lT58+GAylSyzevn07X3/9tS+7Liox56hXDuk+KKp6JPkYvx9ZSmLWOQCahDViePPB1A2u7csu56PmBl6KYiDYHESQOUgCLyGEqISq5HKnw4cP8+effwIwYcIEd/DlYjabeeqpp9xt//7771JdJysri2eeeQZVVRk5cmTZOi0qPVVTSbWlkZSdjKarBJkDShV8pdrSmLN/PlN3zyQx6xzB5mBubTWM+zuMLbfgS9VUsuzZ2DU7QeYgagREEm4Nk+BLCCEqqSr56bxu3ToAjEYjffr08dimY8eO1KhRg3PnzrFq1apSJc2/8847HD9+nGuvvZaRI0fy66+/lqnfovLKduSQbksnR7UTYCpdcrqqa2yK28KqE2vJUW0oQPe6XbmuST8CTWWrkF/UNW1qDugKgeYAgkxB5b6SUgghRNlVyQDMlY/VqFEjQkNDPbZRFIU2bdqwfv16du/eXeJrrFu3jtmzZxMREcFrr73GkSNHytRnUTlpuubM9cotqhpkLl1R1eOpsfx+eClnMhMAaBhSn+EtBtMgpJ5P++ui6Ro5qg10nQBTAEHmIEmuF0KIKqRKBmCnTp0CoEGDBkW2q1fP+cvv5MmTJTp/SkoKzz//PAAvvfQStWrVkgDsEmTL3UooR7WVuqhqhj2DpcdW83fCTsBZFf/6JtfSrU7nUtcJK4qm69jUHHQgwGAl0ByI1WiRwEsIIaqYKhmAJSU5ayiFhYUV2c41Opaamlqi87/yyiskJCQwZMgQbrjhhtJ1UlRamq6Rac8i3Z6BjlaqoqqarhFzZhsrTvxBliMbgK61OzKoaX+CzUHl0Gcdm2pDQyPAYCXIHIjVaJXASwghqqgqGYDl5OQAEBBQdF6NKznf1d4bixcvZtGiRdSsWZOXXnqp9J30kslUtnUQRqMh3/9F0eyqnXR7BllqFlazpURJ6q57fDrjDL/sX8jJ9DgA6gfX4aZWN9AkrKHP+6vnBl6qrhFgsRBsDsJaDbYLkve1/8i99h+51/5TFe51lQzAjMbyqd4dHx/PK6+8AsCrr75KZGT5boZsMChERgb75FxhYYE+Oc+lStd1Mu1ZZGfnYDUbCDdHlTiIybBlMmvHfNYdj0ZHJ8BkZXib6+jb9CqfV5TXdZ0c1YaqqgSbQp1bBpmtl3zgdTF5X/uP3Gv/kXvtP5X5XlfJACww0HlDixvZ8nakzOU///kPycnJjBw5kgEDBpStk17QNJ3U1MwyncNoNBAWFkhqahZqKbfHudQ5NAfptgwy7VmYDCYsRjN2sr0+XtN1tsbvYMnRVWTYnT+vzrUv58bmAwm1hJCe6v0Ia3F0Xceu2bFrDixG54iXSbOSY9fIIctn16ns5H3tP3Kv/Ufutf+U570OCwv0ychalQzAXLldaWlpRbZz5X55M5I1e/Zs1q1bR926dfnPf/5T9k56yeHwzRtDVTWfnetScaGoajp2TcVqsmDAUKJ9HOMy4vn98FJOpDkXctQLrc2wZoNoEtoYoNR7QnpiU+3YNTsWg5lQcyhWoxWjYkRTQaN6/mzlfe0/cq/9R+61/1Tme10lA7BmzZoRHR1NXFxcke3OnDkDXFgNWZTFixe7jymuZtiYMWMA5yrM1atXe9Nl4WeqppJuzyDDnonRYCTIXLI6XNmOHFadWMumuC1o6FgMZgY26cON7a8lPTXHp4GXXXNgV+2YDCbCLWEEmgJkk2whhLjEVckArFWrVgAcP36czMxMgoIKrjrTdd1dL6xdu3Z+7Z+oWGUpqqrrOjvP7mHx0RWk59YG61CjDTc0u44aQRE+DYwubJRtJNQSQpApUAIvIYSoJqpkANa3b19ef/117HY769atY9CgQQXabN++nfPnzwPQu3fvYs/59ddfo6pqoc9v2bKFBx54AIBp06bRqVOnUu8vKcqHpmtk2DJJt2egKEqJi6omZJ5lwZGlHEk5DkCNgCiGNb+eVpEtfNrPizfKDjQHlKoGmRBCiKqrSn7qN27cmM6dO7Nt2zY+/vhjevXqRXDwhdWEdrudKVOmANC6dWuuvvrqYs9ZXKJ+3uetVmu+64mK5yqqmq3mYDVaShTQ2FQba2L/4q/Tm9F0DZPBRL+G19C7wVU+DYxko2whhBAuVXYI57nnnkNRFA4dOsTYsWPZtGkTSUlJbN++nfHjxxMTE4OiKEycODHfKMjOnTsZPHgwgwcP5ocffqjAVyB8QdM10m0ZnM9OxqbZCTQFeh006brO7nP7+PDvr/jz1EY0XeOyyJY83vlBrm3Uy2fBl2yULYQQ4mJV9jdAx44deeWVV3j55ZfZvXs348aNK9Bm8uTJ9O/fP99jWVlZHD16FLhQUV9UTXbVTrotgyw1G7PRXKKA5lzWeRYcWcbBZOcWUxHWcIY2u562NVr7rH+yUbYQQojCVNkADGDUqFF06NCBadOmER0dTVJSEsHBwXTu3Jlx48Z5NfUoqh5d18lyZJNmS0fTS7aVkF21s/bUBtad3IhDVzEqRno3uIq+Da/xWXAkG2ULIYQojqLrul7RnaiuVFXj/PmMMp3DZDIQGRlMUlJGpa114ksOzUG6PX9RVW/tP3+IhUeXcT47GYCWEc0Y1nwQNQNreHW8yWggPCKIlORMj2UoZKNs36lu7+uKJPfaf+Re+0953uuoqODqW4hVVD/5i6o6sJqsGL3clicpO4VFR5ez9/wBAMIsodzQ7Do61Gjjk+BINsoWQghRUhKAiUrPVVQ105GFQTEQZPZuby+HprL+9CbWxP6FXXNgQKFn/R70b9QLq8la5n65NsrWdB2z0VxtNsoWQghRdhKAiUqttEVVDycf5fcjyzibdQ6ApmGNGd58EHWCa5e5T+6NsjUVi8FMmDUIq7H6bZQthBCi9CQAE5VSaYuqpuaksfjYSv45uweAYHMwQ5oOoFOtDmWeEtR1nRyHzTkSh4lQawgBJgm8hBBClJwEYKLSyV9U1YrJi1EvVVPZGLeFVSf+xKbZUFDoUa8rAxv3JdBUsn0gPbGrdnI0lSAsRASEY9LNsm2QEEKIUpMATFQamq6Rac8i3Z6Bjk6QKdCrUatjqbH8fngp8ZkJADQKqc+wFoNpEFL8JuzFsWsObA47ZqOJcGsYNYOjSHNkywomIYQQZSIBmKgUSlNUNd2WwdLjq9mWsBOAQFMgg5pcS9c6nbyuC1aYvBtlh1lDCDQFEGCxeDUaJ4QQQhRHAjBRoUpTVFXTNaLP/M2K42vJVrMB6FanE4OaXEuQOahM/ZGNsoUQQviD/GYRFebioqre5GrFpp3i98NLOZ1xBoB6wXUY3mIIjUMblKkvZd0oW9d1VNWB1DX2DU1TyM42YrPloKpyT8uT3Gv/kXvtP4Xda0UxYDQaK0WdRgnARIXIzh31smsOr1YSZtozWX78D7bEb3NWmjdaGdikLz3qdi3TKkRV17A5bCgKBJmDCDIFYi5Bdf2srAyyszOw2XLQdckL86WzZw1omtxTf5B77T9yr/2nsHttMJgICAgiJCQcg6HiVrFLACb8LtuRTUpOKkCxRVU1XefvhB0sO7aaTEcWAJ1qXc6Qpv0JsYSUug8FN8oOxGK0eH28ruukpiaRlZWG2RxAcHA4ZrMFg0EBKv6b1aXAaFRklMBP5F77j9xr/yl4r3VUVSMnJ4usrHTs9hwiI2tXWBAmAZjwq2xHjjP4UhSsxQQ8p9PPsODIUk6knQKgdlAthjcfTLPwxqW+vq82ys7MTCcrK42wsCiCgkJL3R9ROJPJIKtN/UTutf/IvfYfT/fabIaAgECCgoI5fz6B9PQUwsIiK6Z/FXJVUS25gy8oMvjKdmSz8sSfbIrbgo6OxWBhQOPeXF3vylLX3vLlRtm6rpOVlYbVGiTBlxBCVEFms5WAgGCyszMJDY2okJwwCcCEX1wIvpz7Jh5JOU6aLZ1QSwhNwxphUAzous6OxN0sObaSdHsGAJfXaMuQZgMJt4aV6rrlsVG2qqo4HHZCQiJKfQ4hhBAVy2oNJCsrDVVVMZn8Hw5JACbKXY5qI9WWho7O4ZSjLDyynFRbmvv5MEsovRtcxZ5zBziaehyAGgFRDG8xiJYRzUt1zfLcKNuV1Gk0Sk0wIYSoqoxG5++DilpAJQGYKFc5qo2UnFQ0XeNwylFm7fulQJtUWxqLjq4AwGww0bfhNfRucFWp6m/puo5Ns5fzRtmupE5JthdCiKqrYj/DJQAT5cbmDr5ULEYLC48sL7K9STHxaKf7qREYVeJr6bqOXbPj0FTMBrNslC2EEKJSk99OolzkDb4CTAEcS43NN+3oiUN3kFJMG0/sqp1MRxYKChHWMKICIggyB0rwJYQQotKSETDhczbVTkpOKg5ddVe3T7Ole3Wst+3goo2yLWEEmgJKvUpSCCGE8CcJwIRPeQq+AEK9LJrqTTtPG2XLfo1CCCGqEvmtJXzG7g6+7ASa8le4bxrWiCBToLuavSfhljCahjUq9HnZKLv6WLx4AW+++Uqpj//44y/p0qWbD3vk2a23DuPMmTiGDh3B5Mkv+uSceV/7vHmLqF27jk/OK4SoXOS3l/AJu2onuZDgCyApOwWbZi/yHDc2v85j3lZZN8oWQgghKhv5LSbKzDXypeoOj8GXXXMwe/8vODQHNQNqYNNs+RLywy1h3Nj8OtrXaJPvuIs3yg40BWIpwUbZouq6/vohDBx4ncctW8aOvZ34+DNccUUn3n//Y4/HW63W8u4iAHXr1sNoNBIZWfKVu4UJDg6mQYOGgNSaE+JSJgGYKBO75iAlJxW75iDA5PmX3qIjy4nLiCfIFMS/OtxJqCWEY6mxBSrhu5R1o2xR9ZlMJkwmi8cAzLWLgcFgICgoyN9dy+fTT//P5+fs27c/ffv29/l5hRCViwRgotQuDr48be+zPeEfYuK3oQC3tx7h3lKoeXiTAm19tVG2EEIIUdlJACZKxR18qTYCTAEeg6SEzETmH14CQL9GvWgV6XlbIV9ulF2d6ZpG1oH9OFJSMIWHE9j6MhRD9a2FNnXqV0yf/jVdu3bnpZde47333iImZhNGo5EmTZrxyitvUq9efQBycnJYvHgBGzas4+DBA6SmpmA0GomIiOKKK65g5Mjb6dDh8gLXKCwJ/5FHHmD79r8ZP/4hxo69l3nzfmbZssWcOHEcXddp2rQZgwffwIgRtxTYg66wJPy//97CY489hNFoZO3azRw5coiZM7/j77+3kJycRHh4BF27XsmYMeNo3rxFofdl69YYfvppFgcPHiA5OZk6deowcOD13HnnOObOncNXX31Kp05dSjy6t2vXP/z668/s2LGNc+fOYrFYqFOnLt26defWW0dTv36DQo/9558dzJ//C7t27SQxMQGLxUqrVq254YZhDB58o8fPgpycbBYsmM+aNas4fPgQOTnZREZG0bFjZ0aOvJUrruhU4Ji893DVqvV89dVnLFmygKysLOrVq8+ECY9xzTW93e3j4k4zZ85MNm/eSEJCPCaTiYYNG9O377XceuvoCh+BFVWbBGCixBxeBF85qo1Z++Zh1+y0CG9G/0a9C7TJu1G21WDJ3a+xbBtlV1dpW7eQ+ONMHElJ7sdMkZHUGj2G0K7lvxqwMrPZspk48REOHz7ofiwxMYE6deoCEBt7gieffJS4uFMFjs3KOkVc3CmWL1/KU089x4gRN5fo2jk5OTz++AS2b/873+N79+5m797drFu3lilTPilxrtfatat55ZUXsNls7sfOnk1k2bLFrF69gnff/YArr7yqwHGfffYRs2d/n++x2NgTTJ/+DStXLqdHj54l6ofL/Pm/MGXK2+i67n7Mbrdz5Mhhjhw5zK+/zuW1196hV68++Y5TVZUvvviEH3/8Id/jNpuNbdu2sm3bVtasWcXrr7+DxXIhDeHEiWNMnjyJEyeO5zsuISGeFSuWsmLFUm6//Q4eeWQihkK+hEyZ8jYLFsx3//3YsaM0anRhFfby5Ut4++3X8t3jnJwc9u/fy/79e/ntt3m8996HNG/e0vsbJUQeEoCJEvEm+NJ1nd8OLyEx6yyhlhBubz2iwOpGu5q7bZCPN8qujtK2biHui08LPO5ISnI+PuGRah2E/fPPThRFYcKERxk06EbOnTtLQsIZDAYDqqrywgvPEBd3iuDgYO6/fwI9evQkLCyMs2fPsn79n3z//QyysjL59NMPuO66wSUa9fj559lkZ2czYsTNjBx5G7Vq1eLo0SN89tmH7N27hy1bolmyZCFDh47w+pyapvHqqy9So0YtHnzwYTp37ord7mDFiiV8882X2O123n33TebMmZ8v+Pj55x/dwVf37lfxr389QKNGjTl16iTffz+ddevWcvJkrPc3NteZM3F89NH76LpOjx49ufvuf9GoUSNsNjs7d27js88+5Ny5c7z99qv8/PMCAgMvLNT54YcZ7uCre/erGTfuXzRp0oxz584yZ85M96jk1KlfMWHCowAkJyfz5JOPcuZMHBaLlbFj72HAgOsJDw/nyJHDzJgxla1bo/npp9lYLFYeeuiRAn1WVZUFC+YzcOAg7r9/AgDbtm2hceOmAGzevJHXXvsvuq7TunUb/vWvB2jf/nIcDjtbtkTz9ddfEB9/hieffJTp02cRGRlZ4vsmhARgwmuu4CtHtRFYSPAFEBO/jR2JuzCgMPqymwmxBOd7XtM17LqDcKuzev2lHHjpuo6e5xu0z8+vaSTMnllkm8QfZxLUrn25TEcqlqoxVXzjjcMZM2YcADVr1uSyy5wrbqOjN3L48CEAnnnmPwwYcL37mPDwCFq0aEloaBhTprxNVlYWu3btpHv3giNLhcnOzuaee8YzfvxD7sc6derClCmfcOutw8nMzOCPP1aVKADTdZ3AwCC++moaUVE13I+PHXsvmZmZfP/9dOLiTrNv3x7atesAQGpqKlOnfgVAz569eOutKe5Rt/DwCN5993/897//YcWKpV73w2X9+nXY7XYCAwN56633841U1a07hFq1avPoow+SnJzM1q3R9OrVF3CO2H377TQA+vXrz6uvvu0OGCMiInj++Zew2WysXLmMefN+Yty4+wgKCuL776dx5kwcBoOBt9+eku/n0blzVzp27Mx//zuZP/5YzcyZ3zJ48I00bdqsQL8bN27Ciy++6r4PrpWnqqry3ntvous67dtfzieffJXvNQ0ZMpRu3bpz992jOXs2kenT/48nn3y2xPdNCAnAhFdUTSU1J63Y4OtUepx70+3rm1zrsbBqjsN5jiBTYJX45V1auq4T+/YbZOf+gq8ojqQkDj86oVzOHdCyFY2efb7S/xz79Rvg8fGgoGBuu+0OkpLOF9qmU6cu7j8nJyd5bFMYo9HI6NF3FXg8LCycDh0uJzp6E2fOxJXonOAMAvIGXy7XXNOb77+fDkBcXJw7APvrr7Wkp6ehKAoTJz5TYMpTURSeeOIp1q5dnW/KzRt2u7O9qqqkpCRTq1btfM937tyVt956n7p169Go0YXFN2vXrsFmy8FgMPDYY5M8ThWOGTOOgwf307hxE86fP0dAQACLFi0AoH//6zwGwwaDgUmTnmPDhr+w2Wz89ts8Hn98UoF2vXv38zj1u2nTBvfP5MEH/50v+HKpVas2o0bdyTfffMmSJYt47LFJBXL5hCiOvGNEsVzBV3YxwVeWI5vZ++ah6iptIlvRq0HBD0e75sCgKISYgyv9L22fqA6vsQpo1aq1x8c7duxMx46dCz0uKSmJf/7Z4f67qqolum6jRk0ICfG8vZZr2ionp+QjpG3bti/knBfqkdlsOe4/b968EYCWLVu5Fx5cLDw8gk6duhIdvbFEfXEFqDabjfvvH8dNN91Cz569aNXqMve/8d69+xU4buvWGHefCqv236pVa2bOnOv++8GD+0lPd9YQ7Nv32kL7FBkZ6X4t27dvLfTcnuTN12vWrAWZmZke27l+BllZmRw6dIA2bdoV2h8hPJEATBTJFXxlqtlFBl+6rvPLwQUk5SQTaQ3n1tbDCrTVdR27aifMEoq5GhRUVRSFRs8+X65TkJkH9nP6o/8V267+408S1Poyn1+/qkxBhodHFPm8pmns3LmdvXv3cPLkCU6dOsmJE8dJSIjP1y5vkrk3IiIKv67ZbMk9Z8FaZ6U9r9l84d9V3r66RnTyjkB50qRJ0xIHYG3atOPWW0cxd+4czp5N5JtvvuSbb74kKqoGV17Zg549e9OzZ698uV/gXAgB0LBhY6+vFR9/4efRpEnBacW8mjZ1vpaLf4Yuhb0nTp8+6f7zsGHXedWvhIQECcBEiUkAJgp1cfBlKOIX7frTm9l7/gBGxcjoy272WBE/R83BYrQQZC743KVKURSUcqzKHty+A6bIyHyrHy9miowiuH2HaluSwmg0Fjk9tGVLNO+++wanTxdcBdmoUWMuv7wjixcvKOW1y+cjtqTnTUlJASAgIKDIdhcHSd564omn6dSpC3PnzmHnzu1omsb58+dYtmwxy5YtJjg4mHvvvT/fdGxaWqpXfcorMzPD/efiFkMEBDhfS1aW5/1nLRbP/y4zMjI8Pl6UjIz0Eh8jhARgwqOSBF/HU2NZdmw1ADc0G0jD0IJTHKqmous6IebgSzrp3t8Ug4Fao8d4XAXpUmv0ndU2+CrOrl07mTTpUVRVJTg4mL59+9O2bXuaNm1GixYtCQsLJy7udKkDsMrCFeQUNp3mkp3tOVjxRr9+A+jXbwBJSUnExGxmy5bNxMRsJjExgYyMDD799EOs1gBGjrw1t0+BudfM9voaeQPE4l6L63nXdbzluldRUTX4/fdlJTpWiJKQAEwU4A6+HNkEmosOvjLsGfy4/1c0dK6o2Y4edbt6bJet2gjJLTchfCu0azeY8IiHOmBR1Bp9Z7UuQVGcr7/+AlVVCQkJYdq0mR6LhaakJPu/Yz7WsGFDDh06UGyZidKUobhYZGQk118/mOuvH4yu68TEbOall54nLS2VuXN/dAdgtWvX5tChAx5HHvP65psvCQsLp1u37tStW8/9+PHjR4ssOHvs2BEA6tatW6L+u+rDpaQkk5WVVeiooK7rVWL6XVRe8rVY5KNqKqm2NLLU4oMvTdf46cDvpNrSqBlYg5ta3ODxA8mm2rAYTASbg+QDq5yEdu1Gs3em0PCpZ6l7/0M0fOpZmr3zvgRfxdizZxcA3br1KLRS+5Yt0e4/a1rJ87Uqg86dnV+Mjhw5xJkzZzy2yczMZNs2zwnrRfn44yncccfNvPTS8wWeUxSF7t2vYtCgGwBITEx0P3f55R0BZ2L9+fPnPJ779OlTzJjxDR9/PIWDB/fTvHlL96KGtWvXFNqnpKQkduzYBkD79leU6PVccYVzUYaqqmzY8Feh7X76aRaDBvVl3LjRPglcRfUjAZhw03TNGXw5sgkwWYsMvgD+iF3PoeQjmA0m7rzsZqweNuPWdA2HphJiCcZkkAHX8qQYDAS1aUtYj6sIatNWph294CpDEBt73GNwdezYUX74YYb77w6H3V9d86nrrhtCYGAgmqbx6af/87iY4P/+77NC86WKomk6sbEn2LBhncdARNd1Dh06AECDBheC3EGDbsBoNKKqKl9+6XkKffr0rwHntOA11/TBaDRyww3DAFi9egUxMZs89Efjgw/exW63oygKN9wwtESvp1evPu7VpF999anHEdD4+DPMnPkdGRkZ5OTY3DXEhCgJ+YQWgKfgq+i3xqHko6yO/ROAES2GUCe4tsd2OQ4bgcYArMbyS0QXorRc2/UcPnyI119/iYMHD5CSksyRI4eYMeMbHnjgHtLTLyRYF5d3VFmFhYXxr389CMAff6zm2WefZPfuXaSmpnDw4H5eeeVF5s6d425fkpHqW28dhcViISsriyeeeJglSxZy6tRJkpKS2L17Fy+//B93aYebb77dfVzt2nXcxXEXL17Aiy9OZu/e3aSkJLNv315ee+2/LFmyEIBx4+5zj3zdc894ateug6ZpPPvsJGbM+IbY2BOkpqawY8c2nnrqcVavXgHA6NF3FVqyozBWq9VdN+z06VOMHz+O5cuXcPbsWfdWR4888iDnz59DURQee+xJGdkXpSJDEsIdfGXYM72qTJ+Sk8pPB+ajA93qdKJzbc9D/HbNgaIoBFsk8V5UThMmPMqOHds4f/4cy5cvYfnyJQXaDBp0A7t3/8PJk7HExlbdqabRo8cQG3uc33//lQ0b1rFhw7p8z7dufRlms4Xdu/8p0d6UDRs2YvLkF3nzzVc4cyaON9542WO7W265nWHDbsr32H33PUhychK///4ra9asZM2alQWOGzHiZu666x7338PCwvnww894+uknOHXqpLvsxcVGjRrDgw/+2+vXkdfAgYNISUnh44+nEBd3ildffbFAG6PRyOOPP0XPnr1KdQ0hJACr5jRdI82W7nXwpWoqcw7MJ8OeSd3g2gxtdr3HdnruRtthllAs1aDml6ia6tdvwLRpP/Dtt9PYtGkDiYnxGAxGatSoQdu27Rk+fCTdunXn/fff5uTJWDZsWIfD4aiSVc8VReGZZ/7DVVddw/z5c9m/fy9ZWVnUrVuPwYNvYNSoMUye7Bz5KaxEQ2Guv34ILVu2Zu7cH9m+/W/i48+gaRpRUTW4/PKODB8+ki5dCuYjGo1GnnnmP/Tvfx2//vozu3b9Q0pKMkFBwbRr156bb77dY4DTuHFTvvvuR+bP/4U//ljN0aNHsNls1K5dmyuu6MSIEbfQvn2H0t2oXLfccjvdu1/Fzz/PZsuWaBIS4tE0jZo1a9G165Xcdtto2YhblImil7SyoPAZVdU4f77kNWfyMpkMREYGk5SUgcNRsgThvMGXN9OOAEuPrWLdqU1YjRYe7ngfNQOjPLbLduRgMpiItIZjNHj/bboyc93r+PjzJCScpkaNeu5imsL3TCZDid/TonRc9/qBB+5hz55d3HjjcJ577r8V3a1Lkryv/ae4e2232zh3Lq7En+VRUcEYjWWf1al6X+OET+QNvqxeBl97zx1g3Sln0uvNLYcVGnw5a35phJiDL5ngS4iqbNOmDaxevYJGjZpw113jPOYsZWdnc/z4UcBZEV8IUb4kAKuGLg6+jF4EX+ezk5l70FmMsme9K+lQs02hbXNUO8FS80uISsNoNLqLybZv38HjdOCcOTPdVeBdixOEEOVHArBqpjTBl0NzMHvfL2Sr2TQKbcCgpgMKbWtTbZgMRqn5JUQl0qlTF+rUqUt8/Blefvk/3Hffg3TteiXBwSHEx59h2bJF7lWQgwffWOhG1UII35EArBopTfAFsPjoSk5nnCHQFMjoy0ZiKmRaUdN1VE0j3BomNb+EqETMZjOvvPImTz/9BOfPn+O999702K537748+eQzfu6dENWT/JasJjRdI92W4Qy+jBavg6+dibvZfMZZHfu21sOJsIYX2jZHzSbAGECAh4KsQoiK1aHDFcyaNZeffprNxo3rOXXqJKrqoEaNmrRq1ZohQ4bSq1dfGbkWwk8kAKsGdF0nw5Z5IfjyMjE+MfMsvx5aBEC/htdwWWThS64dmgMFA8GWIKn5JUQlFRkZxYMP/rtAfSxZmSeE/8lvykucruuk2zJIs6djMZq9Dr5sqo1Z++dh0+w0C2tC/8Z9iryGLTfx3iKJ90IIIUSxJAC7hOm6TrrdFXx5P/Kl6zq/H15KQmYiIeZgRl12U5FTls7Nts0EmQJ91XUhhBDikiYB2CXKHXzZnMFXYYnznmyJ3862xH9QUBh12UhCLSGFtlV1DQ2dEIvU/BJCCCG8JQHYJejika+SBF+n08+w8MgyAK5r0pfm4U2KbJ/jyCHIFCibbQshhBAlIAHYJShbzSHdnoHFYC5R8JXtyObH/fNw6CqtI1vQu0HPItvbVDtGxUiQ1PwSQgghSkQCsEuQrmug6yWqxaXrOvMOLeJcdhIR1jBuazUcQxFBlabrODQHoZZgzFLzSwghhCgRCcAEABvjYth9bh9GxcDoy24myBxUZPscNYdAYwABpgA/9VAIIYS4dEgAJjiRdoolx1YBMKTpQBqFNiiyvbPmF1LzSwghhCgl+e1ZzWXaM/lx3zw0XaNDjbZcVa/gJr15uWt+mYKl5pcQQghRSlU+eWf37t1MnTqV6OhokpOTiYyMpEuXLowbN44uXbqU6pyaprFw4UJ+//139uzZQ2pqKkFBQbRu3ZohQ4Zw2223YbFU/eBD03V+Pvg7KbZUagREMbLljcUm09tUG2aDmSCz1PwSQgghSqtKB2BLly5l0qRJOBwO92MJCQksXbqU5cuXM2nSJMaPH1+ic6anpzNhwgSio6PzPZ6SkkJMTAwxMTHMnTuXr776itq1a/vkdVSUtSfXcyDpMCaDiTvb3FzsHo6uml9hUvNLCCGEKJMqOwW5a9cunn76aRwOB926dWPWrFls3LiR2bNn06NHDzRN4/333+ePP/4o0XknT55MdHQ0iqJw11138dtvv7Fp0yZ+/vln7rzzTgwGA3v27OGRRx5B06ru3mlHko+x6sSfAAxrPoi6wXWKPcbmyCHQFECA1PwSQgghyqTKjoB99NFH2Gw2WrVqxbRp07BanUFBVFQUU6dO5d577yUmJoZ3332XPn36YDAUH2vu3LmTFStWADBx4kQefPBB93ORkZFcccUVtG7dmpdffpkdO3awfPlyBg8eXD4vsByl2tKYc2A+Ojpdal9Btzqdij3GrtoxKEaCzcFS80sIcclRVRWjUUb2hf9UyRGww4cP8+efztGbCRMmuIMvF7PZzFNPPeVu+/fff3t13qVLlwIQFhbGvffe67HN6NGjqVPHOVq0du3aUvW/Iqm6xk/755Nuz6BOUG2GNS8+gNR0HbvmIMQcJDW/RIU6ffoUU6d+xYQJ93HjjQPo27cHgwf349577+Sjj6Zw5Mjhiu5itbd48QJ69epGr17dSEiIdz8eF3fa/fiyZYtLfN5HHnmAXr268cYbL/uwt2Cz2Zg69Stmzfq+wHNvvPEyvXp1Y9Som3x6TSGgio6ArVu3DgCj0UifPn08tunYsSM1atTg3LlzrFq1im7dil7dB3D27FnMZjNt2rQpNMleURQaNWpEfHw8CQkJpX8RFWTVibUcTT2BxWDhjstuxmI0F3tMjpqD1WghUBLvRQXJzMzg888/YcGCX1FVNd9z6enpHDx4gIMHD/DLL3O4+ebbeeSRJzCZquTHm/Czxx9/iH/+2cm9995f0V0R1UyV/ITau3cvAI0aNSI0NNRjG0VRaNOmDevXr2f37t1enffdd9/lnXfeISMjo8h2sbGxgHOkrCrZd/4ga09uAGBkyxupFVSj2GMcmgo6hFhCpOaXqBDJyck88cTDHDp0AIDu3a9m6NARtGnTlpCQEBITE9m7dxc//PAdJ0+eYO7cH4mPj+ONN97zKvVA+IfJZKJBg4YABAUVXejZnxITEwt9LjIyigYNGlK3bj0/9khUF1UyADt16hQADRoUXTC0Xj3nP5qTJ096fW5FUQgJCSn0+TVr1hAf7xxW79q1q9fnrWhJ2SnMPfg7AD3qduWKWu2KPUbXdXIcNkIsQVil5peoAJqm8cILz3Do0AEUReGpp55jxIib87UJCwunRYuWDB48lJdf/g9//LGKdevW8tNPsxg9+q4K6rm4WK1atZkzZ35Fd6NEHn74MR5++LGK7oa4RFXJr4dJSUlA8SNQrtGx1NRUn1w3NTWVN954w33tYcOG+eS85c2hqfy4fx5ZjmwahNTjhmYDvTrOptmxGM2EmIPLuYdCeLZo0e9s3+7M4Zww4dECwVdeJpOJF154hdq1nTmaM2ZMJScn2y/9FEKIkqqSI2A5OTkABAQUvQ+hKznf1b6s13z00Ufd04+TJk0iPDy8zOc1mcoWAxuNhnz/BzDqBoyqAVPuY4uOLudk+mkCTQHc1e5WAszFj2Zpuoai6UQEhmI1FZ8nVh14uteViaZrHEo+SmpOKmHWMFpGNKuy08auhbazZn0HQN269bjttjuKPS4gIIDRo8fw5Zef0bJlK06fPk2zZs0BmDr1K6ZP/5quXbvz0kuv8d57bxETswmj0UiTJs145ZU3qVevvvtc69b9wcKFv7F37x5SU1MICQmldes2DB58A9ddN7jQ1cC7dv3Dr7/+zI4d2zh37iwWi4U6derSrVt3br11NPXrFxy513Wd1atXsHTpIvbt25t7vRAaN25Cz559uPnmWwkOLnxk3pN33nmDBQt+JSwsnN9/X1ZoTtzy5Yt59dX/oigKP//8e77ptlOnTjJ//i9s3RpDfHwc6enpBAUFU79+A666qie33XYHERERXvUnLu40t902HIAXX3yVQYNuKNBm+/a/mTNnFvv37yUlJZl69RowZMiNjBo1ptjzl7Svb7zxMkuWLHT/ffr0r5k+/WsA/vprS742DRo09Dh6l5OTzYIF81mzZhWHDx8iJyebyMgoOnbszMiRt3LFFZ3ytVcU2Lp1C//+9wMYjUbWrt3MkSOHmDnzO/7+ewvJyUmEh0fQteuVjBkzjubNWxT7ui/mcDhYvHgBq1Yt5+DBA2RmZhASEkqLFi3p0+dahg27qdAc55ycHFasWMLSpYs5ceI4qakpREXVoEuXbowefRctW7byeNzJk7H8/PNsYmI2Ex9/BqPRRP36Dbj66msYNWqMx/eI694OHTqCMWPG8d57b7Jr1z9YrVZatmzFO+98kG+q+q+//mTRot/Ys2cXqamphISE0q5de4YNG0mvXgVzwV3/PBUFdL3oe2Y0KmX+XVwaVTIA8/dS4aysLB555BE2bdoEwIgRIxg9enSZz2swKERG+mZ0KSzsQoK82QZ6lp1gSxBbT//DhtMxANzb5Xaa1a1f2CnyybRlEWGOIiowQspOXCQkJACDwVBh/2g9+Tv+H37aN5+knBT3Y5HWcG5vcxNd6lxegT0rvYMHDxAbewKAG28cRmCgd/Xnbr31dm67bRRmc/4vDgaD831ss+Xw5JOPcOjQQfdziYkJNGhQH4PBQFZWFi+88Czr1/+V7/jk5CSiozcSHb2R/2/vzONqSv84/rm3um3aVHbJdrNXtpRM9i1ZJoNBshuGIVsYjH03DGMdRbKVKEuIFEIUIpI1rUq3fe9u5/fH/d3j3u5SkVvxvF8vL7dznuec53yf55zzPc93eS5dCsCWLTtkPsL8/f2wbdtmUBJPfB6Ph7i494iLew9/fz9s2rQNvXo5SNVbvfpPXL9+VWpbbm4unj+PxvPn0fD3P4v9+w+jSZOmFZIBADg6OuLSJX/k5eXiyZMI2NnZyy0XHHwdAGBpaYUmTT4rh+fP+2Hnzm0QCPhS5fPz8/D6dR5ev47FlSuXcOiQJxo1+vxcEcsZEH1giu8RyQ8XJlP23jl8+AA8Pf+T2hYfH4cDB/biwYN74PG4AERuImXrfklblT3XxMcXl2EwZD+WExLisXTpQiQkxEttT0//hBs3ruHGjWsYN248/vhjoUJfxLCwUPz115/gcrn0towMDoKCriAk5AZ27PgHNjY9FLazLDweDwsXzsWTJ4+ktufkZOPx40g8fhyJixfPY9++QzA0NJIqk5r6Ee7ui/DmzWuZ67l2LRA3bgRhxYpVcHSUtvz4+/th164dUtcAlOLduzd49+4NLlw4h40bt6JbNxupemLZ5uRk448/ZtG+eFxuKXg8LvT1RR8cpaWlWLt2FUJCgqXqZ2dn4d69MNy7F4aBAwdj5co1chVLZR/MQiEDTCYTBgY65U7ofAtqpQKmrS1SNsqb2aroTJkysrKyMGfOHERFRQEA+vbtS5shvxahkEJeXtFXHUNNjQl9fW3k5RVDIBBCSAnxnBOLtMJ0qDPUEfBOFO7du6kdmmk1Q25O+efjCXgQQgiWljZySr+ufd8TYlkXFJRAKBRCIKDA51d/Mt6n6c/x3wvZEPrs0lwceuaFGR1cYFWvdilhDAYQHf2M/tvSsnOFZc38f6qUsuWFQpFS9Pz5MzAYDMyePQ+DBjkiMzMD6elpEApFPmd//umO+/dFytfgwY4YPXosGjVqjPT0dFy8eB7+/n6IinqMpUsX4p9/DtAzS2lpqfj77+2gKAo2NnaYNGkqmjZtCi6Xh+joKOzbtxuZmZnYuHEtzp69RD/Hbty4RitfY8b8CkfHETAxMUFeXh5CQ4Ph4XEI6emfsGPHVuzYsafCMmzf3hINGzZCaupHBAVdQ/fudjJl8vJyEREh+rAcOHAILbOYmBfYvl2kSLZv3xFTpsxA8+YtoK6ujqSkRPj4nEJY2C1wOOk4fPgAVq5cKyNncR+IjykQCKXKSPbP5csBtPLVpUt3TJ8+C2ZmzfDxYwqOH/dEWNjnlD8UJV33S9u6ePFyLFiwBC4uY/DpUxpcXKbAxWUK3W7xuUT/S4+nnJwczJ//O9LSUsFiacLFZTL69RsIAwMDxMW9x7FjHnj8OAJnzpyCujoLv/02F8DnWRmRDIRYs2YljI1NMWvWHFhbdwGPx8eNG1dx5MhB8Hg8bNmyAT4+ARUOJjlxQjSTpqamhqlTZ6J3734wNDREVlYWLl8OgI/PKbx//w4HDx7A4sXLJPqJDze3PxAfHwcNDQ24uExB//6DoKenh9jYl9i//x/Ex3/A5s3r0bIlm54JCwkJpj84mjUzx4wZs9GpkxX4fD4ePgzHf/8dQFZWJhYtWoBDhzzRurWFVD8CopktFouF5ctXw9a2J5KSksDn82h5r1+/BiEhwWAwGBgxwhkjRoxC/foNkZGRjitXLsPX9xSuX78GTU0tuLuvpI/PYIie2QKBUOEMmEBAQSgUIje3CMXFAvmF5KCvr10llpBaqYCJfbvy8/OVlhP7fhkZGSktp4j4+HjMnDkTCQkJAERflFu3bpX5sv4aquoFLhAI8ejjM5x9exE5ErMgAGCqbYK+TR3AF5R/LiFFoYTPhT5LDwxKrUYoGDUNQQXkKIaiKHCFvG/WFiElhO+bC0rLnH17ERZ1W38TcySLqfFNZkgpSmTWENO0qVmVHt/RcTgmTHAFAJiYmMDCog0A4O7d27Ty5eo6DTNmzKbr6OsbYOFCdzRpYoY9e3bi2bMoXL4cgJEjRwMA7t0LA4/Hg7a2NjZv3iH1Nd6gwRCYmtbDvHmzkJOTg8ePI2BvL5oFu307FIBI8fjjj0V0HQMDQ0yaNBV8Ph+enocREfEA+fn5CiO/y8JgMDBw4BB4eXng7t3b4HK5MjMEoaE3wefzwWKx0Lv3Z9/QM2dOgKIoGBsbY+fOvVKBScbGJrC0tMasWVPw8uULREY+rFB7FFFaWoJDh/YDALp06YadO/fQSq2BgSE2bdqB1auXIzQ0WG79L20ri8UCi8Wix6+6unqFozO9vT2RlpYKJpOJLVt2onv3z7NU1tZdYGlpjdWrl+HWrRCcPOmFwYMdYW7eXEoRoCgK2to6OHTIE3Xrfo5Id3GZgqKiInh7H0Vq6ke8evUS7dp1qFC7xGNpyJBhcHWdRm83MDDEvHkLkZ+fjytXLiE09AYWLXKnr93P7wzi4+MAAOvWbUavXr3punZ29mCzLeDqOg65ubk4c+YEVq5ci9LSUuzZsxMURaF58xY4cMBTSvZOTiPRuXNXzJjhiry8XOzYsQWHDh2V2+5Jk6bC0VFknpaURWTkQwQHBwEA5s1biDFjPrsh6OvrY+7cBWjcuAl27tyCS5cCMGLEz2jTpt3/5Qup/5VRXR/TtVIBa968OSIiIpCamqq0XFpaGoDP0ZCV4cmTJ5g9ezZycnIAAJMmTcLy5ctrbFj7k0/yZ0EAgFOcgVdZb9DeuE25x+EKSsFSY5HFtqsAiqLw95P9iMtNqNZ25JTmYvGd1d/k2C0MzLGw8+xvooRJpoMxMDCs0mP37t1P7vbLl0XKbL169TF16ky5ZcaM+RXXrgXizZtXCAg4TytgYhOZQCBAbm4OTE2l14q1tu6CzZt3oEGDhmjatBm9XVwvLy8HfD5fxldr1Khf0LZtOzRq1KTS6RsGDRoKLy8PFBYWIjz8Hhwc+kjtv3lTZH60te0pFdTUsaMldHV10bq1hdyocAaDASsra7x8+QI5OdmValNZHj9+hOzsLACiQIuy189gMDB//iLcuRMqkwNO1W0FRDNXgYGXAAB9+w6QUr7EMJlMLFq0HPfv3wWXy8WFC+cxf/4imXJDhgyTUjjE9OzZC97eImUlNTW1wgqYeCxlZWXJ3T9lygwMGDAIDRtK+yGKzdDdutlIKV9iTExMMWKEMyIjH9L34oMH95GRITIbzpu3UK7sGzdugqlTZ2D37h2IiXmOt29fS82CienTR35g2IUL5wAADRs2wujRY+WWGTnSGT4+p5CcnIhLlwJoBaw2UCsVsNatRdOfCQkJKCoqkvtQoiiKzhfWrl3lOuTu3buYM2cOSktLwWQy4e7ujsmTJ391u78VQqEQvq8ClJYJjLuBtnXZSmdBBEIBKAqoo6Fba523ax7Ef+5LkfQl4vN5Ch2Hv4TWrdlyt4sjLu3tf1Lqa9qnT3+8efMK79+/RV5eHvT19WFl1RmAKLP6jBmuGDnSGXZ29mjd2oJWUOW93CwtO+PevTC8ffsGs2ZNgZPTCPTo0ZN2hjcyMoKtrXz/rfIwM2uGtm3bITb2JUJCrkspYBkZGfT1Dh4s7RAvOdNQFoqi8OFDHFJSROl95ClFleHxY5GPqoGBgcKXp4mJKdq0aYeYmOcy+1TZVgB4//4tCgpE1peyCq0kRkZGsLLqgoiIcDx9+lhumbZt2yuoW5f+zeVWPIjM0tIa79+/w/37YZg/fzYGD3aEjY0treQ1bNhIKtAEECUyfv1a9K60s+ul8NgzZ87BzJlz6L/FY0dHRxddu3ZXWK9Pn/7YvXsHACAq6omMAqapqQkzs2byquLpU5HrT+vWFigpURzR3L59ByQnJyI6+qnCMjWRWqmAOTg4YMOGDeDxeAgLC8OgQYNkyjx9+pT+CujVS/GgKktUVBTmzp2L0tJSaGhoYPv27RgyZEiVtf1bEJvxTsr5Wh653DzE5yWhhYH8gQ4AJQIu6mjoQEudLLZdFTAYDCzsPPubmiDf5cRh/zPPcsvNsZyKVoYtqvz838oECUDKSTgnJwc6OlWXDkXejFpRUSEKCgoAAM2aNVda39zcHIDoBZ+RkQ59fX20adMOo0ePhZ+fDzIyODhy5CCOHDmIunWN0a2bDezsesHOzp72/RLj7PwLbt26iZcvX+D161j6ZdismTlsbGxhb+8AK6vOUrPvfD6/jNOzNJIfpYMGDUVs7Evcv38XJSUltE9sSMgNCIVC1Kmjh5495T8j8/LyEBn5EB8+vEdKSjKSk5MQH/8BxcVV5xsqXq6oUaMmSss1a2YuVwFTZVsB0HkgRW0qf5xERIRLLckkiaIIUkk3F6oiNrT/M2XKDDx8GI6UlGTa6Z7BYKBVq9bo0aMn7O0d0L699GxaRgaHPkdlgjzS00UWJjOzZkotQ8bGJqhTRw8FBfly5aCvbyD3GVJUVEjPWN65E4qBA0PLbROHU7tWp6mVCpiZmRmsra0RFRWFPXv2wN7eHrq6nx/OPB4PO3fuBACw2WzY2tpW6Li5ublwc3NDcXEx1NXVsX//foVLHdUksouVK19i8rkFCvdxBVywmOrQ1ag5Gaq/BxgMxjdNYtu2LhuGmgYyfn+SGGkalDv7WRNp1syc/p2Skiw3fYMilC2srKamJjclg6TJszxTn5bWZyWqqKiY/r1gwRJYWXWGn58PoqOfQigUIisrE0FBVxAUdAW6urqYMmWGVIJYTU0t7Nv3H86ePYPAwAt0VF1CQjwSEuLh63saDRs2xpIly2lz1/XrV7Fp02fH97KI0ygAQL9+g7B37y4UFxfj3r076NdvIIDP5sc+ffqDxWJJ+cAIBAIcOXIQPj4nZRQ9FosFK6vOYDAYiIqSP7NTGcSzSeUFSylKw6HKtgIixUBMRcdJcXGx3P1qalX7CjYyqoujR0/ixAkvBAVdwadPaaAoil6qy9v7KFq1YmPZslVo06YtAFEgxuf2VjxgTXy/lP2gkIe2tjYKCvLlKsOKZrbLW5GmqupUJ7VSAQOA5cuXY+zYsXj37h1cXFywdOlSWFhYICEhAbt27UJkpEjzd3Nzk9Kuo6OjsXTpUgDAxIkTMXHi5wfhwYMHab+ymTNnokuXLko7VE1NrVpCV8tipF2xfGR6LPkPMCElBF8ogJGWAdTJYtu1CiaDiV9aD1fo/wcAo1sPr3XKFyAyp4h59ChCJoxdEXl5efjlFye0b98JTk4jFPqXlEVb+/PLtKhI+ayJ5P6yL6Devfuhd+9+yM7ORmTkQzx69BCRkQ/B4aSjsLAQ//67G5qaWhg1ajRdR0NDA+PHu2D8eBckJSUiIuIBHj2KwOPHkSgqKkRqagrc3d1w+LCXQvOpIoyMjGBjY4f798Nw8+YN9Os3EKmpH+nZpEGDZGf4//lnB86fPwsAMDdvDnt7B7Rs2Qrm5i1gbt4cGhoa8PA4VCVKjdj3rKREvpIiRuzfVJ1tBaT7u6LjRFJh/9bo6OjS5sJ3797SY+np0yfgckXpIRYsmIMTJ87CxMRE6nqUmfnKIr5fFCmXkoiV1srIQVPz87t1wgRXzJ49r8J1awu19m1raWmJtWvXYs2aNYiJiYGrq6tMmWXLlqFv375S24qLi/HhwwcAnzPqA6KUFWfOnKH/3r9/P/bv36+0Dd27d4e3t+IXn6poa9IKRpoGSs2QBix9mOvLn14u5XOhra4FTTVieqyNWNXriBkdXGQiYI00DTC69fBal4JCTKNGjWBh0RavX8ciJCQY06f/VqEI5NDQYBQWFiIiIlxugkZF6Orq0qaShIQPSsuKI8YAoH79BnLLGBkZYeDAwRg4cDAoikJk5EP89dcK5Ofnwc/vjJQCJknTpmZo2tQMzs5jwOVyERBwDnv3/g0ej4cLF85h8eLlGDrUCUOHVnwljkGDhuL+/TA8fHgfpaUlCAm5AUCU4FZS0QVE6TT8/f0AAD/91AcbNmyVa2LKzc2p8PmVIZZfUlIiKIpSaNL++DFFZpuq2wpAKlFtQsIHpclSxeOkQQP5Y+Rb06pVa7Rq1Rrjx7ugqKgIXl4eOHnSCwUF+bh27TImTpwMU9P6dHl5MhYTF/cOwcHX0ahRY/TvP4iWQ1JSAoRCoUIzpPjDA0Cl1tTU09ODrq4uCgsLkZb2UWlZZeOmJlP7PoslGDt2LPz8/DBs2DDUq1cPGhoaMDQ0RJ8+fXDs2LFKOc6/efOm3K+ZmgqTycSYNiOVlnFsMUDuLAhPyAeTwYAucbyv1VjV64j1dssx33oWprT7FfOtZ2Gd3fJaq3yJETtYp6amwNf3VLnlxeH7AFCnTh307y/rH6oIBoNBZy6/e/eOUoft27dDAADm5i3o6K89e3bi119/xl9/rZB77O7de9DZ38VJJ0tLRUlhf/7ZkZ7FkYTFYmHMmF/RokWr/9f7Mh8Xe/ufoKuri9LSUjx6FEkntZSX0T82Nob2CRo2bITcFytFUXjy5POMklD45SH83buLXEQKCgoUzlIVFRXi+fNnMturoq2VfXG3aNGK7nNx2gd5ZGdn49kzkRN5+/adKnWOL+HTpzTMmzcLTk4DER5+V2a/jo4OZs+eR7vriCMYDQ0N6TQvERHhCo8fGnoTx497YteubVBXV4elpRUAkdnv0aMIhfVu3Qqhf3foUPHnEYPBQMeOlgBEM+DKlhWbNm0iRowYjPXrV1X4+DWBWv/Gbd++PXbu3ImwsDC8ePECDx8+xMGDBxX6fdnY2OD169d4/fo15s37PKXZsWNHentF/9WE2S8xneuLZkEMNaXNkQYsfYxv4yw3BQVFUeAJeNDR0AFLjSw3VNthMphgG7VE1wbWYBu1/C4U6gEDBqNDB9HL6/Dh/QgIOKewbFFRIVauXIq0NJEbgavr9HLXiy2Lk9MIACLHcE/Pw3LLnDvnQ2cLF+cuAkTJRZOSEnH/fphUDjMxFEXh3bs3AIDGjUX+bJqamsjMzER6+idcuHBebnLpvLw8fPqU+v96yh3VFaGpqUmn3rhw4Rzt6D9woKz5UdIvKT5e/kzgiRPHpGYBebwvDzSxsupMp0X499/dcs1ghw7tk2vqqoq2io/B50tn0VeEmpoaPfsYEnIDkZEPZMoIhULs2rUNPB4PDAYDQ4cOq9CxvwZjYxPExb1DdnYW/Px85Trvp6Qk0xMNkkEPQ4eKxvH9+3dlsugDoqzz4nvvp5/6QF1dHXZ2vejoyr17/0ZhoayPcVpaKo4dEyXYbdmylcKoT0U4OY0EIPLPPnDgX7llLlw4jzdvXiMzM0MqvUttoPY/oQk04lmQ3zq6YmTLoZjWYSIWd/1dYf6vUgEXLDUWcbwn1FiYTCbWrt2E+vUbQCAQYMeOzZg7dyaCg4OQkBCPvLxcxMd/wNmzZ+DiMpbO6t6nT3+MG1f++oFlsbd3oEPxvbw8sGHDX/9fmzEPcXHvsHv3Djqkvn37jvjll89Lko0ePRYsFgvFxcVYsGAOrl69jJSUZGRnZyMm5gXWrPmTDt3/+ecxdL3x410AiNIbuLn9jocPw8HhpIPDSUd4+D0sXDgXBQUFUFNTw/DhihcjLw/x7Js40SybbUGvkylJp05W9Dq6R48eRkDAOaSlpSIzMwOPHkVg5cqlOHRon1Sdr7EeqKur01nZ37x5hd9/n4HIyIfIzc3B+/fvsHHjGpw75ys3qKIq2ipeTio8/B4+fUqjc5IpY/Lk6ahXrz6EQiHc3Rfh2LEjSEpKRF5eLp49i8LixfNpM++4cRMrrXh8Cerq6vR6qQ8f3seyZQvx9OkTZGVlIi0tFaGhwVi06A9QFAVdXV0p37/Ro8fCzKwZKIqCu7sbTp8+gdTUj8jIyMCdO7cwd+5MZGdnQUtLC9OmzQIgmp1dsGAJAODDhzjMmjUFt2+HIDs7CxkZHAQGXsSsWVOQm5sLDQ0NrFixptLX9NNPfdCjh2gFBz+/M1i5cilevIhGbm4OPnyIw6FD+/D331sBAI0aNVaakqQmUmt9wAjyYTKYaGnYHCbaxtBWkkxVlPNLSHJ+EWo89es3wMGDnli/fjWePHmEp0+f0IpMWdTU1DBu3ETMnDnni3xCGAwG1qzZgNWrl+PBg/u4di0Q164FypTr2rU7/vpro1Q0ZZMmTbFs2Sps2rQWaWmp2LhxjdxzODuPob/sAZFiJF7AOzr6KRYtknU2ZrFYWLZsFczNlac9UIa1dRfUr98Anz6J0gfIm/0CRCapOXPmY9eubSguLsaOHZtlyujo6GLkyJ9x6pTICpCcnPjFK44AgI2NLZYtW4Xt2zfh9etYuLn9LrWfzW6D5s1bICjoSpW3tWvX7oiJeY53797A2Vk0UxUQcA0mJiYK26uvb4Ddu/dhyZIFSElJplOOlGXs2AmYNet3OUf4NkycOBmvX8ciLOw2vU5iWXR1dbFp0w6pVCza2trYvv0fLF78B5KSErFv327s27dbqp6Oji7Wr98ilaqib9/+yM1dht27tyM+/gP+/HOpzPkMDQ2xevUGerWJysBgMLB27SasWrUcERHhuHUrRMqkKaZhw8bYseOfKk1VowqIAvaDUirgQVdD55umSCAQqgpT03rYs+cgHj2KwM2b1/Hy5QukpaWipKQE2traaNy4Kbp06Qonp1FfvWyRjo4utm//B7dvh+DKlct49eol8vPzYGRUF61atcbQoU746ac+cv2NBg4cglat2PDzO4OnT5/g06c0CIVC1K1rjI4dLTF8+Ch07txVpt6iRe6ws7PH5csBiI19iezsLKirq6Nevfro1s0Go0ePq1SOJnmIlyby9j4KNTU19O8/WGFZZ+cxaNrUDD4+pxAbG4PCwgJoaWmhceMm6NatB5ydx8DQ0AgXLwagoCAft2+H0v46X8qwYSPQrl17nDrljWfPopCZmQETE1P06zcQrq7TsHfv39+kra6u01BSUoKbN68jJycbBgYGSEv7qFQBAwAzM3McP34GAQHncOtWCD58iAOXy0W9evXQqZMVRoxwlsm59a1RV1fHpk07EBwchKCgK3jz5jVyc3OgpaWFBg0awda2J375ZRyMjWWvrXHjJjh27BT8/f0QEhKMxMR4lJSUoF69+rCxscOECZPkOtGPGjUaXbp0g6/vaTx6FAEO5xM0NFho3LgJHBz6wMlp1Fcp57q6dbBz5x7cuROKq1cvIzb2JXJzc8BiaaJZM3M4OPSBs/PYCqXDqGkwqMpkeSNUKQKBEFlZX5e3RF2dCSMjXWRnF9J5fIp4RcgtzVM4A8YVcAEwUFfLkKSdqARiWX/6lIX09I8wNm4IDQ2iwH4r1NWZZC1SFUFkrTqIrFVHebLm8bjIzEyt9LO8bl3dKlmMm9iefjDEOb/qaOgS5YtAIBAIhGqCKGA/GKWCUmiraZHlhggEAoFAqEaIAvYDwRfywQATuizieE8gEAgEQnVC3sI/CBRFoVTAhS7J+UUgEAgEQrVDFLAfhFIBF5pMFnTUa1+kCIFAIBAI3xtEAfsBEFBCCCkh6rB0ocaUTWZIIBAIBAJBtRAF7AeglF/6/5xfxPGeQCAQCISaAFHAvnO4Ah7UmerQ0dCplavFEwgEAoHwPUIUsO8YIUWBL+SjjoYONEjOLwKBQCAQagxEAfuOKRWU/D/nl1Z1N4VAIBAIBIIERAH7ThFQQjDAgC5Lh+T8+maQVbwIBAKh9lK9z3DyZv5OYYABXXVdsMhi21WOeBFmoZCs50YgEAi1FaFQpIBVl380UcC+UzTVNaGjYDFuwtehpqYOBoMJLre0uptCIBAIhC+ExysFg8GEmlr1+EgTBew7REONBX2WHsn59Y1gMBjQ0tJBSUkhKIqYIQkEAqE2UlpaAhZLk8yAEaoODaY6WW7oG6OtXQcCgQC5uZlECSMQCIRaRlFRPni8Emhp6VZbG0huAgLhC2CxNGFoaIKcnAzweKXQ0tIBi6X1f/8wkm+tKhAKGRAIiHKrCoisVQeRteqQlTUFgUCA4uJClJYWQUdHD9raRAEjEGodWlo6qFu3PoqLC1BUVIDCwrzqbtJ3BZPJJIEOKoLIWnUQWasORbJWV9eAnl5d6OjUqYZWSbSjWs9OINRyWCxNsFia0NevC4FAAIoiD9aqQE2NAQMDHeTmFpHZgm8MkbXqILJWHYpkzWQywWSq1YiVYYgCRiBUAQwGA+rq5HaqKtTVmdDS0kJxsQB8PlFqvyVE1qqDyFp11AZZEyd8AoFAIBAIBBVDFDACgUAgEAgEFUMUMAKBQCAQCAQVQxQwAoFAIBAIBBVDFDACgUAgEAgEFUMUMAKBQCAQCAQVQxQwAoFAIBAIBBVDFDACgUAgEAgEFUMUMAKBQCAQCAQVw6AoiqyHUE1QFAWh8OvFr6bGhEBQMzP9fm8QWasOImvVQWStOoisVce3kjWTyaiSpYyIAkYgEAgEAoGgYogJkkAgEAgEAkHFEAWMQCAQCAQCQcUQBYxAIBAIBAJBxRAFjEAgEAgEAkHFEAWMQCAQCAQCQcUQBYxAIBAIBAJBxRAFjEAgEAgEAkHFEAWMQCAQCAQCQcUQBYxAIBAIBAJBxRAFjEAgEAgEAkHFEAWMQCAQCAQCQcUQBYxAIBAIBAJBxRAFjEAgEAgEAkHFqFd3AwiVJyYmBh4eHoiIiEBOTg6MjIzQuXNnuLq6onPnztXdvBpJaGgo/Pz8EB0djezsbGhpacHc3BwDBgzAhAkTUKdOHbn1CgoK4OHhgevXryMpKQksFgvNmzfHyJEjMXbsWKirK76FhEIhzp07h3PnzuHNmzcQCARo1KgR+vfvj6lTp8LIyOhbXW6NgsPhwMnJCdnZ2di2bRtGjBghtxyRdeXIz8+Hl5cXgoODkZiYCD6fj6ZNm9LXbGBgoLAuh8PB4cOHcfv2bXz8+BG6urpgs9n45ZdfMHz4cKXn5XK5OHHiBC5fvoy4uDgwGAw0bdoUQ4cOhaurK7S1tav6UmsE9+/fx4kTJ/Ds2TPk5uaiTp06aN++PcaOHYuBAwcqrEfGdcWYM2cObt68iZMnT6Jr164Ky1Xn2L1+/TpOnTqFmJgYlJSUoH79+nBwcMDUqVPRuHHjSl8zg6IoqtK1CNXGtWvXsGjRIvD5fJl9TCYTixYtwvTp06uhZTUTPp+PpUuXIjAwUGGZJk2a4PDhw2jZsqXU9szMTIwfPx7x8fFy61lbW+PIkSNylTehUIj58+fj+vXrcuuamprCw8MDFhYWFb+YWsqsWbNw69YtAFCogBFZV46YmBjMnDkTGRkZcvc3bNgQXl5eaNasmcy+Dx8+YPz48cjKypJbd8CAAdi9e7dcxaC4uBhTpkxBVFSU3LrNmzeHl5cX6tevX4mrqfls3boVnp6eCvcPHz4cW7duBZMpbVQi47pinDx5EuvWraN/K1LAqnPsbty4EcePH5e7T09PD3v37oWtra3c/QqhCLWG58+fUx06dKDYbDY1fvx46tGjR1RmZib1+PFjysXFhWKz2ZSFhQUVGhpa3U2tMWzatIlis9kUm82m3NzcqGfPnlFZWVlUbGwstXv3bqp9+/YUm82m+vbtSxUWFtL1BAIBNXr0aIrNZlPdunWjfH19qfT0dCoxMZHavXs31a5dO4rNZlN//PGH3PNu2bKFYrPZVJs2bai///6bSkhIoDgcDnXu3Dmqe/fuFJvNpvr06UMVFRWpShTVgo+PDy1/NptNBQQEyJQhsq4caWlpVLdu3Sg2m011796d8vX1pT5+/Ei9e/eO+vfff+kx7eTkRPH5fKm6+fn5VN++fSk2m005ODhQ165dozIzM6l3795Ra9asoftp69atcs89b948is1mUx07dqQ8PDyojx8/UqmpqdTRo0cpS0tLis1mU6NHj6aEQqEqRKES/Pz8aLmMGzeOevDgAcXhcKinT59Ss2bNovft3btXqh4Z1xXD19eXsrCwoOUYGRkpt1x1jl1vb2/6+KtWraLevn1LZWZmUtevX6f69OlDsdlsqmvXrlRaWlqlrp0oYLWI6dOnU2w2m3J0dKRKSkqk9nG5XGrChAkUm82mhgwZQgkEgmpqZc0hLS2NfsitWrVKbplbt27RN9aRI0fo7YGBgUofCKdPn6b3R0VFSe1LTU2lX4L79u2TqRsbG0sr0ocOHfq6i6zBJCYmUlZWVuUqYETWlWP+/PkUm82munTpQsXGxsrs9/X1peUVEhIite/w4cMUm82mOnToQMXHx8vU3bFjB8Vms6n27dtTKSkpUvuePXumtB/v3LlD7798+fJXXmXNYeDAgRSbzaaGDRtGlZaWSu0TCoXUpEmTKDabTVlZWUntJ+NaOaWlpdTatWulng/KFLDqGrtFRUWUjY0NxWazqRUrVsjUTU1NpXr06KH0PaMI4oRfS3j//j3u3LkDAJg9ezY0NTWl9mtoaGDx4sV02SdPnqi8jTWN4OBg8Pl8MBgM/PHHH3LLODg4wNraGgBoMxkAeHl5AQDs7e3lToePHTsWLVq0AACcPXtWat/JkyfB4/Ggr6+PqVOnytRt06YNRo0aJbfu94JQKIS7uzuKioroa1UEkXXF4XA4CAoKAiDymWnTpo1MmVGjRsHAwAAsFgsxMTH0doqiaBPKqFGj5Jon58yZAwMDA/B4PAQEBEjtO3bsGACgRYsWcn1tevXqBXt7ewDfh6wBkZ+d2Hw4fPhwsFgsqf0MBgNjxowBABQVFeHdu3f0PjKuFXPjxg0MGzYMJ0+eBAC0b99eafnqHLsXLlxAdnY2mEym3PdIgwYNMGXKFADApUuXUFpaqvRaJCEKWC0hLCwMAKCmpoaffvpJbhlLS0sYGxsDAG7evKmyttVU0tPToampiUaNGsHExERhOTMzM7o8AOTk5ODZs2cAgL59+8qtw2Aw0Lt3bwBASEiI1L7bt28DAGxtbaGlpSW3vvi4iYmJePv2bQWvqPbg4eGBx48fo3379vjtt98UliOyrhxBQUEQCoXQ0dHB+PHj5ZZRV1fHnTt38Pz5c8ydO5feHhsbS49xRbLW1tZGjx49AEg/QyiKop9BvXv3BoPBkFtffNyIiAjk5+dX8upqHpLXKRAI5JaR9DcS+4CRca2YvLw8zJ07FwkJCdDR0cHq1avh7u6utE51jl1xX7Rr106hf1ifPn0AiJTw8PBwpdciCVHAagmxsbEAgKZNm0JPT09uGQaDQX8RS375/qi4ubkhOjoaFy9eVFouMTERAOiosVevXoH6f2xK27ZtFdZr164dACArKwsfP34EAPB4PLx//15qvzwkj/u99dXr16+xZ88esFgsbNmyRWmUF5F15YiOjgYAWFlZybyYJQNz5L20xc8QoGKyfv36NX3M5ORk5OXlSe2Xh/i4AoEAr169UnottYE6deqgefPmAIDAwEBwuVyZMv7+/gAAIyMjuiwZ18pRU1PDiBEjEBgYiAkTJpRbvjrHrvi3srqtWrWChoYGAODFixdKr0USooDVElJSUgCg3FDXhg0bAhANOoIIRSkmANGN+vTpUwBAly5dAHyWNaBc3mJZA5/lnZaWRt/4yuqamprSN+z31FdcLhdLly4Fl8vFvHnzwGazlZYnsq4c4pkOsQnm+vXrmDRpEqysrNChQwf06dMHW7duRU5OjkxdsaxZLBbq1aun8BxiWfN4PHz69EmqLqBc1o0aNaJ/13ZZi1m4cCGYTCbevHmDadOmITIyEpmZmXjx4gUWLFiA0NBQMBgMuLu7064hZFwrRltbG9evX8e2bdukxosyqmvsCgQCpKWllVuXwWDQ55Y8X3mQPGC1hOzsbACAvr6+0nLi2TGxxk9QDJfLxerVq0FRFNTV1WlfDrGsASjNpSQ5EymWt2RdZX3FZDKhq6uLnJwc5ObmfvE11DT27t2LV69ewdraGtOmTSu3PJF15eBwOABEslqxYgXOnTsntf/jx4/w9PTE1atX4eHhIZVaRSwvPT09hWYY8X4xeXl5aNy4cYX7SfJjp7bLWszAgQOxd+9ebNu2DREREZg4caLU/ubNm8Pd3Z02QwFkXCtDQ0MDTZo0qVSd6hq7ubm5EAqFACr+7q1MX5AZsFqC2LFPkT+AGPEXWGUcAX9EhEIh/vzzT3r2a+rUqbT5QFJ2ZYMdJJHsC3EdSRPFj9ZXT548gYeHB7S0tLB582aoqamVW4fIunIUFhYCAJ2c087ODr6+vnj+/DnCwsKwePFiaGhoIDU1FbNnz0ZRURFdt6LPEHmy/pJ+kmeuq60UFBRAR0dH7r5Pnz4hMjISxcXF9DYyrquW6hq7knW/RV8QBayWUJGXGaFiCAQC/Pnnn7RvmI2NDebPn0/v/xpZl03E+KNQVFQEd3d3CAQCLFq0iFZmy4PIunKUlJQAEM2EOTg44MiRI7C0tKRNMzNmzMD27dsBAAkJCfDx8aHrfo2sf+Tnz/r16+Hu7o5Xr17B1dUVQUFBeP78OUJDQ7Fw4ULw+Xx4eHhgxowZtBJGxnXVUl1j91v3BenpWoJ4iYTytOuKfin8qBQXF2PevHk4f/48AFHk6P79+6UcxSWXo1D2FS9+GQKfv34kv5J/pL7asmULEhMTYWNjAxcXlwrXI7KuHJLtd3d3l/tyGTJkCB3WL05ZAVT8GSIpa/H5JPtJWX15/VSbES8/BADLli3DihUrYG5uDhaLhUaNGmHWrFk4fPgwmEwmIiMj6VQJZFxXLdU1dr91XxAFrJYgti+XF9ot9iP4ntYIqyoyMjLg4uJChyh3794dnp6eMk76krZ+ZfKW3CeWt6QPQkFBgcK6QqGQNifV9r66c+cOfHx8oKuri82bNyv10SgLkXXl0NXVBQCYmJjILJ0liTjvlDiaDvgsa2WyAqT9R8XykuwnZfXl9VNtxtfXF4BoubJJkybJLWNra4sBAwYAAM6cOQOAjOuqprrGrq6uLv2R8y3evUQBqyWITTqpqalKy4kjNiSjawiiF9HYsWPx/PlzAMCgQYPg4eEhN0LS3Nyc/i0ODZeHZF+II2gaNmxIfwEpq8vhcMDj8eg6tZkrV64AEPkn9e3bFxYWFlL/+vXrR5ddunQpvR0gsq4sYufl8maXxONa8qtdLOuSkhKFa+kBn58hGhoadF7Br+mn2ow4CauVlZVSc1T37t0BiK6/uLiYjOsqprrGLpPJRNOmTWX2l4WiqC969xIFrJbQunVrACK/DknHWkkoiqLzpSjLWfKj8erVK4wfP54OLZ4yZQp2794tk9VaTOvWrelZHGW5jF6+fAkAMDQ0pG86JpNJz0xI5q5RVBf4sfuKyLpyiPP8paenKzWJZGZmAoBUyL74GQJUTF5sNps2zderVw+GhoYAKtZPTCbzu1gkWqzgiP9XhHgMUxQFHo9HxnUVU51jV5xKR9l53759S4+RyvQFUcBqCQ4ODgBEDwJxVt+yPH36lP466NWrl8raVpOJj4/HlClTkJOTAwaDgRUrVmDZsmVKv2br1KmDzp07A5DNUC2Goih66aKyshavVHD//n2FL8nQ0FAAomUsJB8utZF169bhyZMnCv8FBgbSZTds2EBvB4isK4vkc0DRahcURdHZuC0tLentbDabfskrknVxcTEePHgAQLGsFdUFPsvayspKYcLo2oTY8vDo0SOpRLdlefToEQBRbi59fX0yrquY6hy74rovXryg08AoqqulpYVu3bqVez1iiAJWSzAzM6PXLNyzZw9t+xfD4/Gwc+dOAKLBamtrq/I21jS4XC4WLlxIK6Xr16+Hq6trheqOHDkSgGh9SHlLS/j4+CAuLg4AZI7p5OQEJpOJrKwsHDx4UKbuq1ev6OzZkyZNqpTPVE2ExWJBV1dX4T9Jp1TJsmKIrCtOr169aPPIzp07pXIciTl58iQSEhIAQGodTgaDgREjRgAA/Pz85C5fs3//fuTm5kJDQ0NmqSNxP7158wZ+fn4ydcPCwnD37l0Asv1UWxk6dCgA0Yziv//+K7dMZGQkHezg6OhIbyfjuuqozrE7cOBA6OjogMfjYceOHTJ109LS6LUmnZ2dlSb+lqFSS3cTqpWnT59SFhYWFJvNpkaNGkWFh4dTWVlZVFRUFDVp0iSKzWZTFhYW1M2bN6u7qTUCb29veoV7d3d3qqCgQOm/oqIiui6fz6ecnJwoNptNWVlZUd7e3lRaWhqVnJxM7dmzh2rXrh3FZrOpefPmyT33mjVr6HOvX7+eiouLozIyMih/f3/KxsaGYrPZVN++fan8/HxViaPaSEpKomUREBAgs5/IunLcunWLfg4MGDCACgwMpNLT06mEhARq586dVJs2bSg2m03NmTNHpm5+fj7Vs2dPis1mU7a2tlRAQACVkZFBxcXFUWvXrqXluGXLFrnnnjFjBsVms6n27dtTe/fupZKTk6m0tDTKy8uLsrKyothsNjV69GhKIBB8azGoBKFQSE2cOJGWy8KFC6nHjx9TWVlZVHx8PHXo0CGqU6dOFJvNphwcHKisrCy6LhnXFefBgwf0tUZGRsotU51j97///pMaA7GxsVRmZiZ148YNqm/fvhSbzaa6d+9Offz4sVLXzaCo/y9YRagV+Pj4YM2aNXR23rIsX74ckydPVm2jaigDBgyg13msCI0bN5aaok5KSoKrq6vCpSWsra1x9OhRqTBnMSUlJZgzZw7u3bsnt27dunVx6tSpCufLqs0kJyfTjvjbtm2jv2QlIbKuHBcuXMCqVasUmqdsbGywd+9euZm/o6OjMX36dIUZuwcOHIh//vlHrpk+OzsbU6ZMUegPY25ujlOnTtEO0N8Dubm5mDdvHh4+fKiwTLNmzXDgwAGZyFQyrivGw4cP6SjTkydP0lG8ZamusSsUCrFixQp61rEs2tra8PDwoJezqyhEAauFxMTEwNPTExEREcjOzoauri6sra3h6upKTI//Jysrq9KyKKuAAaKwZQ8PD9y4cQPJycmgKAotWrTAsGHD4OLiotCRHxDdtOfOncP58+fx9u1blJSUoGHDhujduzdmzpwJU1PTL7q22kZFFDCAyLqyJCUl4dixYwgLC0NaWhq0tbXRsmVL/Pzzzxg5cqTSRdAzMjJw+PBh3Lp1C6mpqVBXV4eFhQWcnZ3h7Oys1EeSy+XC29sbgYGB+PDhA/h8Ppo2bYqBAwdi+vTplTPB1BKEQiGuXr2KgIAAxMTEIDc3F7q6umjZsiUGDRqEsWPHylWiADKuK0JFFTCgesduUFAQfHx8EBMTg8LCQpiYmKBnz56YOXMmvT5rZSAKGIFAIBAIBIKKIU74BAKBQCAQCCqGKGAEAoFAIBAIKoYoYAQCgUAgEAgqhihgBAKBQCAQCCqGKGAEAoFAIBAIKoYoYAQCgUAgEAgqhihgBAKBQCAQCCqGKGAEAoFAIBAIKoYoYAQCgUAgEAgqhihgBAKBQCAQCCqGKGAEAoFAIBAIKkbxaq0EQhXi4uKCiIgIAMCCBQswe/bsCtVbt24dTp48CQC4efMmmjRp8s3a+K0QXzuLxcLz58+ruzk1Bh6PhyNHjuDSpUtITU0Fk8mEqakpNm/eDGtra4X1LCwsvvrcr1+//upjVIZly5bB398fAHD37t0qXURZcrHzsWPHYt26dVV27Ori9u3buHnzJh4/foyMjAwUFhbC0NAQ9erVg42NDQYMGIDOnTtXdzNrBOL7wd7eHh4eHtXcGkJlIAoYQeXs378fAwcORMuWLau7KYRqZMmSJbh69arUtoKCAtSrV6+aWkSobhITE7F8+XI8evRIZh+HwwGHw0FMTAw8PT3Rp08frF27FvXr16+GlhIIXw9RwAgqh8vl4s8//8SpU6fAZBIr+I/I+/fvaeWrcePGWLRoEZo0aYLc3Fw0btxYad2AgACF+0aOHAkAMDU1xX///VdVzSWogIyMDEyZMgXJyclgMBgYMGAA+vbti6ZNm0JTUxP5+fl4+fIlzp8/j/fv3yM0NBSJiYk4ceIE6tatW93NJxAqDVHACNVCVFQUjh8/jsmTJ1d3UwjVwLt37+jfv/32GxwdHStct23btuWWYbFYFSqnKrZs2YItW7Z8k2M3adJE5SbVb8GuXbuQnJwMJpOJvXv3on///jJl7Ozs4OrqitWrV9OK2O7du78Lsyvhx4NMPxBUCpPJhLq6SO/fvXs3kpKSqrlFhOqguLiY/l0b/foIVUtxcTEuXLgAABg8eLBc5UuMhoYG1q5dS8+U+vv7Iz8/XyXtJBCqEqKAEVSKuro6pk2bBkD00F25cmU1t4hQHQiFQvq3mppaNbaEUBNISEgAj8cDAJibm5dbnsVi0bOmXC4XHz58+JbNIxC+CcQESVA5c+fOxY0bNxAXF4cHDx7A19cXY8aMqfRx9u7di3///RcA4OPjAysrK7nllEVSiiMUBw0ahD179uDx48fw8vLCkydPkJeXB1NTU/Tp0wczZ86kncMTExPh4eGBsLAwpKenQ09PD126dMGsWbPQsWPHctsdGRkJDw8PREVFobi4GA0bNoSDgwMmT56MRo0aKa1bXFyM06dPIzg4GHFxcSgoKIChoSE6dOgAJycnDBkyRK5f3cOHDzFp0iQAIh+q9PR07Nq1C+/fv4euri4sLCywbdu2Sjk0c7lcBAQEICgoCLGxscjLy4Oenh5atWqF/v37Y+zYsdDS0pKqIxkNK0bcLgA4fvw4bGxsKtyGr0E8fvT09BAZGQlPT0+cPHkS6enpMDY2ho2NDbZt20aXFwgEuHr1KkJCQhAdHY2srCxwuVzo6enB3Nwc9vb2GD9+PIyMjGTOpSwKsm/fvkhJScGkSZPw559/Ijw8HKdPn0ZUVBSys7NhaGiIzp07Y/z48ejRo4fMsZVFQZ4/fx7Lly8HIBp3TCYTx44dw40bN5CUlASKotCsWTMMGDAArq6uqFOnjkJ5FRQU4MSJEwgKCsKHDx/AYDDQrFkzODo6wtXVFTdu3MDChQsBfF3E8uPHjytUbuTIkWCz2ahbt65Cpa0q+qx9+/Y4f/48Xr9+jaNHjyI8PBxZWVkwMTGBjY0NZs2ahebNmwMQ+bF5eHggJCQEqamp0NLSQseOHTFt2jTY2dkpPEfz5s1x7do1xMTE4NChQ3j8+DFyc3NhYmKCHj16YOrUqWCz2RUXYhlycnJw4sQJ3Lp1C4mJiSguLoaxsTEsLS3x888/w8HBQWn969ev4+LFi7QMtbS0UK9ePXTv3h1jxoxBu3btvrhtPzJEASOoHBaLhQ0bNmDixIkQCoXYtm0bHBwcqj2a6eDBg9i9ezcoiqK3JScnw9vbG8HBwfDx8cHLly+xaNEiFBYW0mWysrJw48YN3Lp1CwcOHECvXr0UnuPYsWPYsmWL1Dni4+MRHx8PHx8fbNmyBUOGDJFbNzo6GnPnzsWnT5+ktnM4HISGhiI0NBTe3t7Ys2eP0kjC27dv459//qFnobhcLlJSUioVffjq1SssWLBAZuYhKysLERERiIiIwNGjR7F3794KKaXVzc6dO6Wc9tPS0qRm5pKSkjB79my8fftWpm5WVhaysrLw5MkTnDx5EkePHv3iVBnbt2/HkSNHpLZxOBwEBQUhKCgIs2fPxoIFC77o2B8+fMCCBQvw8eNHqe2xsbGIjY3F2bNnceLECbmKU1JSEqZOnYrExES5dS9evIjRo0d/UbsA0ayXjo4OioqK8PDhQ2zZsgVz585VqhC2bNlSaSR1VfaZv78/Vq1aRc/SAcDHjx/h7++PkJAQHD9+HFwuF7Nnz0ZGRgZdprS0FHfv3sW9e/ewYcMGpTK6evUqlixZInWO1NRU+Pv749KlS9i0aRNGjBihsL4ibt++jcWLFyMvL09qe2pqKlJTU3Ht2jX07dsX27dvl5E3j8fDggULEBwcLLM9Pz8f79+/x+nTpzFz5kwsWrSo0m370SEKGKFa6NKlCyZMmABvb2/k5+fjr7/+wsGDB6utPREREQgKCkL9+vUxbdo0dOjQARwOBwcPHkRsbCxSU1OxePFiREdHQ1NTE25ubujWrRu4XC78/Pxw+fJl8Hg8rF27FtevX5c7C8XlcrF582ZoaWlh8uTJ6NWrF/h8PoKDg3H69GmUlJRg0aJFaNSoESwtLaXqvn37Fq6urigqKoKuri7Gjx8PW1tb6OnpISUlBZcvX0ZwcDCioqIwbdo0+Pj4QEdHR+61/vPPP9DT08P8+fPRtm1bvHnzBiwWCwwGo0Kyio+Px5QpU5CVlQUA6NOnD0aOHIlGjRohPT0dly9fxtWrV5GamopJkybBx8eH/nrfsGEDioqKcPPmTezdu5fe1qFDBwCAmZlZxTqsCiksLMR///2H1q1bY+7cuTAxMcGDBw/w008/ARD12/Tp0xEfHw8AGDRoEIYOHYr69eujsLAQ79+/h5eXF1JSUpCZmYnVq1fDx8en0u24evUqOBwOGjZsiMmTJ8PS0hIlJSW4fv06Tp8+DYqicODAAfTu3VvhbK8yfv/9d3A4HAwbNgyOjo4wNjbG+/fvcejQIcTHxyM1NRVr1qyRUQDz8/MxceJEpKWl0dc/atQoGBkZ4fXr1zhy5AjevHmD7du3V7pNYrS0tDBu3Dh4enoCAI4ePQofHx84ODigV69e6N69O5o2bVrh41VlnyUlJWHlypXQ0dHB77//jm7dutEzSuHh4cjNzcXKlSuRkpKCoqIizJw5E/b29lBTU8O1a9dw4sQJUBSFTZs2YfDgwXKVyoyMDCxbtgx8Ph+jR4+Go6MjWCwWwsLCcPToUZSWlsLd3R2GhoblzlZJEh4ejtmzZ0MgEMDY2BgTJkxA165doaWlhfj4ePj5+SEiIgIhISGYO3cuPDw8pD48Dh8+TCtf/fv3x8iRI9GgQQMUFBQgOjoanp6eyMnJweHDh2FtbY2+fftWuG0EooARqpGFCxciJCQEKSkpCA0NxaVLl+Dk5FQtbcnOzka9evVw9uxZqZk4Gxsb9O7dGyUlJYiIiIC+vj58fHxokwMA2NragsfjISgoCElJSXjz5g3atGkj9zw6Ojrw8vJCp06d6G09evSAnZ0d5syZA4FAgHXr1uHcuXNS9ZYsWYKioiI0aNAA3t7eUopKp06dMGTIEJw8eRLr1q3DmzdvsH//fixevFhuG4RCIXbt2oWePXsCQKUTWq5du5ZWvpYtW4YpU6ZI7e/fvz8cHBywbNkyFBUVYfHixbhw4QJtsgJEMydizMzMqjViUSgUQk9PD8ePH6fTGXTt2pXef+HCBfpFLjYTSmJvb4/Ro0fDyckJKSkpePr0KdLS0tCgQYNKtYPD4aB169Y4efIkDAwM6O22trZo1KgRduzYAUA0G/MlChiHw8GGDRvwyy+/0NssLS3Rv39/DB06FBwOB3fv3gWHw5Eyke7du5dWvhYtWoSZM2fS+6ysrODo6Ijp06cjKiqq0m2SxM3NDa9evcL9+/cBAEVFRbh69SqdrqRBgwbo1q0b7O3t4eDgINdsKKYq+ywvLw/a2to4deoUWrduTW93cHBA3759kZ6ejufPn0NDQwNeXl7o0qULXaZr167Q0NCAp6cnCgsL8eDBA7kBBuIggu3bt2P48OFS9R0cHODq6goul4uNGzfCzs4OGhoa5YkTJSUlWLJkCQQCAdhsNry8vKTSdVhaWmLEiBH0rGt4eDh8fX3x66+/0mXEzyE7Ozvs27dP6vi2trZwcHCAs7Mz+Hw+fH19iQJWSYgTPqHa0NHRwfr16+m/N27cSL/Yq4OZM2fKmEGNjIzQvXt3+u9JkyZJKV9iJB88CQkJCs8xd+5cKeVLsv6oUaMAAC9evJDKmH/v3j1aYXF3d1c4SzRhwgS6radPn5YyZUhiZmZGK1+VJTY2ln5B9u7dW0b5EjNq1Cj8/PPPAERZ5+/cufNF51MVgwYNUphLKjk5GY0aNYKWlpbCFRx0dXVpPywAMmbiirJkyRIp5UvMmDFj6BnKL0050bFjRynlS4y+vj5t9qYoSur4paWl8PX1BQBYW1tLKV9i6tSpgx07dtDRzV8Ki8XCkSNH4ObmBl1dXZn9aWlpuHTpEtzd3dGrVy8sW7ZMytwnSVX32bhx46SUL0AUjSlZf9iwYVLKlxjJMmVNuJKMHDlSSvkS07lzZzpwKSEhgb7/yuPixYvgcDgARM9WRePbzc2N/jA6fvy41D6xfMX7y9KmTRvMnj0bv/32W7V9PNdmiAJGqFZ69uxJv6izs7OlFDJVI89JFoCUUmZrayu3jLGxMf27qKhIbhkmk6nUB0ScRBQQKV1ibt26Rf8uT3ESmycKCgoULntU1rxZGcLCwujfY8eOVVpW8ktasl5NRNmMkpubG0JDQxEVFaU04afkrBGXy610G9TV1eU62QOAgYEBrZhJ+h9WBmVjR1Kplzx+eHg4nTJEfJ/Ko0mTJl+s1EuipqaG3377DWFhYdixYwccHR3lypzH48Hf3x+DBw+WmzW/qvtM0X0v+WxQ1HeS51f0bABEH1CKcHZ2pn+HhoYqLCeJ+Lmhp6cn96NPjLq6Ot13cXFxUopoixYtAAB+fn44duyY3HQfc+fOhZubW6Vy+RFEEBMkodpZvnw5wsLCwOFwcOXKFTg6OirNA/StUJSBncVi0b8VreEnWUbSwV4Sc3NzubMbYiQjid68eUP/ljTXSc7GlUdSUpJc82LDhg0rfIyySDo0l2cGa9euHTQ0NMDj8Wp8otCKmAvFfn0CgQBpaWlITExEQkIC3r59i2fPniEmJoYuK5lmo6IYGxtDU1NT4X4dHR3k5OSAz+dX+tiA8nxrkv6CAoGA/i059sR+eoro1KkTbt++/UVtK4uuri6cnJzg5OQEiqLw9u1bRERE4P79+wgPD6cVmfz8fPz222/w9/eX6yNWVX2mSHaS972iIJaKPBs0NTXRvn17hedv2rQpDAwMkJubi7i4OIXlJBH3XX5+fqWCQpKSkmjFcubMmXBzcwOPx8PmzZuxfft2WFlZwdbWFj179kSnTp1IGpmvgChghGpHX18ff/31F+bOnQsAWLNmDbp37w59fX2VtUFNTU0mZYKicl+K5CyZPPT09KCurg4+n4/c3Fx6e3Z29hedr2zUkxhlkWXlkZOTA0D0YlPmgwOIvqwNDQ3B4XCkrqcmUp5M+Hw+zp07B39/f8TExMidLfnaZbUUBU2IEZsgFb3Ev+b4kgEYksfPzMykf5fX3+WN7y+FwWCAzWaDzWZj4sSJKC4uho+PD3bv3o3i4mLk5+fjyJEjWLt2rVS9quwzeSbRsnzts6G8+kZGRsjNzVVodi1LVTw3hg4diqKiImzbtg25ubng8/l49OgRHj16hL1798LQ0BD9+vWDq6vrF0f+/sgQBYxQIxgwYAAGDx6Ma9eugcPhYMuWLdi0aVOVHLsisxE17StO0slWPONhZGSEo0ePVvgYitJ6VDTaUR6VffmLZV/T1/xUJpPs7GzMmDFDyqSrpqaGZs2aoUWLFmjbti26du2K6Oho7Ny5UxXNVRmSSkt599HXjCtxWggmk0mbvRShra2NyZMnw8LCgl7KLDw8XKpMVffZt34+VOT44pnJijjgA5+fGxYWFti6dWuF21J2tk8clRkSEoKbN2/i3r179IdYTk4Ozp07h4CAAKxcuRLjx4+v8HkIRAEj1CBWr16NBw8e0Dd1eT4Fir7ay1JTlikpbxZI0rwkOdtgaGgIQOQ/YmFhUa3KjNiEKhQKkZ2drdS/hsfj0V/TykyvNZ21a9fSL3J7e3tMnz4d1tbWMjOmkZGR1dG8b4rkOMzKylK6UPqXzrhwuVzax8rS0pJ2+i8PW1tbtG3bFrGxsTIO9LWtzyoyQyyWryI3iLKIZ59LS0u/OspYW1sbjo6OcHR0BEVRePXqFe7du4fr16/j2bNnEAgE2LhxI3r16lWpdCE/OjX7s5TwQ2FsbExn7QaAVatWSa0ZWBbJr8aSkhKF5VJTU6umgV9JQkKCUkff6Oho+rekP5g4+qq0tFTKJ0ced+7cgaenJ4KCgr5JRKmkmeHZs2dKy8bExNCRmOXNatRUMjMzERQUBEDkw3fo0CHY2trKNVeXTXD6PSD54n758qXSspL+VJWBxWLRK0C8fPmywiY24LNZVXK2tzb2WV5eHlJSUhTuF696AUBhipuyiJ8bCQkJ5T4LAgMD4eXlheDgYKlAgYyMDEREREh9xDIYDLRt2xbTp0+Hr68vPQvJ5/NrfLBNTYMoYIQaxciRI+kEmCkpKbh06ZLCspI+YsnJyXLLZGVlSSk21UlpaSlu3rypcL+fnx8A0QPO3t6e3i75+9SpUwrrCwQCrF27Flu3bsUff/yhVHn9UiTbUl6y0dOnT9O/qyJCrjpISkqiTW8WFhYKUy3k5eVJRatKOrLXZuzs7GjFRdm9mJmZ+VWpRsRBNzweD3v27KlQHQ6HQyt9ksEptbXPxIuRy0MyL+DAgQMrdDzxvUpRFM6cOaOwXH5+PlatWoVNmzZhyZIltGXhwoUL6NmzJ1xcXGiFVh6SiWG/JPr3R4YoYIQax7p162inV0W5rADp2RgfHx+Zsnw+H2vWrFF6DFWzceNGuV/d/v7+9EOud+/eUrnG+vfvT0/rnzt3DhcvXpR77M2bN9OKaL9+/ZSai76Udu3a0S+70NBQmbxBYgICAhAQEABANPvVu3fvKm+LKpA0wUVFRdGzEJIUFBRg8eLFUrMM38uLSF9fn06PEhkZCW9vb5kyXC4Xy5YtQ2lp6RefZ/r06dDT0wMgupfFKyYoIisrC3/88QdKSkqgrq6OqVOn0vtqa58dPnwYL168kNn+4MEDeHl5ARDlBFOWUkKSX375hQ4uOXDggIyfHCByJVi2bBmdemT06NHQ1tYGIPpoEvubHThwQKGJWVIxLy9SliAN8QEj1DgaNmyIxYsXy0Q1lcXa2hpmZmZITEzE8+fPMXnyZLi6usLU1BRxcXE4efIkYmJi6DLVjba2NjgcDpydnTFjxgxYWVmhqKgIgYGB9ELNhoaGWLNmjVQ9dXV1bN26Fa6uruDxeFi6dClCQ0Ph6OgIU1NTpKSkwMfHBw8ePAAg8rdasWLFN7uOjRs3wtnZGXl5edi4cSPCw8MxcuRINGzYEBwOB4GBgbhy5QoAUXj9rl27vjpJZ3XRrFkztGvXDi9fvkR6ejrGjx+PqVOnolmzZigsLER0dDR8fX1lzNw1xe+wKnBzc0NISAjS09OxYcMGREVFYfjw4ahbty7ev3+PY8eO4dWrV1J1KuuQX79+fRw8eBDTpk1DSUkJvL29cfXqVQwePBjdunWjUzx8+vQJERERuHDhAq00rFu3TsrEXVv7rLi4GC4uLpg8eTJ69uwJgUCAW7duwdvbGzweD5qamlILrZeHvr4+1q9fDzc3N3ppJmdnZ/Tr1w/6+vqIj4+Ht7c3PYvYpEkTzJs3j65vYmICFxcXeHp6Ijk5GcOHD4erqyvatGkDXV1dep1K8cxnjx49pFaQIJRP7XwqEr57fv31V1y5ckWpk6yamhq2b9+O6dOnIz8/nw6PlqRfv34YOXKk1IOluhBHXh0+fFhuVFLDhg1x6NAhuTmpunTpgsOHD8PNzQ05OTm4cuUKreRI0qBBA+zbt09pzqevxczMDN7e3vj999+RnJyMkJAQhISEyJRr2rQpdu3aVWGflZrKli1bMGnSJOTk5OD169dwd3eXKWNqaoo5c+bQHw3v3r1TdTO/GYaGhjh69CimTZuGtLQ0BAYGIjAwUKqMtbU16tevj2vXrgGQzn1VUbp27Uovp/Xs2TNkZGTgxIkTOHHihNzypqamWLVqFQYNGiSzrzb22YwZM3DkyBHs378f+/fvl9pnbGyM/fv3y2TjL4+hQ4eCz+dj9erVdPoOea4DrVu3xoEDB2RS/7i5uSElJQVBQUFIT09XuN5nly5dKmw6JnyGKGCEGgmDwcCGDRswYsQIpQ72VlZWuHr1Kjw9PREaGoqPHz9CW1sbFhYW+OWXX+Dk5FThzNGqYNGiRbC0tISXlxdiY2MhEAhgZmaGwYMHY9KkSUrzDdnZ2eHmzZs4ffo0bt26hffv3yM/Px86Ojpo1aoV+vXrh3Hjxn1Vnq+K0qZNG1y9ehVnz57FjRs38Pr1a+Tn58PY2BjNmzenF3wWmzNqMxYWFrh48SKOHDmCsLAwpKSkgKIo6Ovro2XLlujTpw9Gjx4NPT09/Pfff/j48SOuXbuGBQsWfFVqhppEq1atEBgYiGPHjiE4OBiJiYkQCoVo0aIFRowYgQkTJmDlypV0+fJymimiQ4cO9Gzu7du38ejRI2RkZCA7OxsMBgOmpqZo2bIlBgwYgEGDBikc67Wxz1xdXfHTTz/hyJEjePr0Kfh8Ppo1a4ZBgwbh119//eJI4uHDh8POzg4nT55EWFgYEhMTUVhYiDp16qBNmzYYMmQIfv75Z7lKM4vFwp49exAaGoqAgAC8ePECHA4HFEXB2NiYXod28ODB381YVyUM6kuz+hEIBAKB8H9+++03hIaGQldXF0+ePKnu5tQKli1bRrsf3L17t8IpJgjfB2QGjEAgEAhyyc3NxcKFC2FmZoahQ4eiW7ducsuVlJTg6dOnACqeJoFA+NEhChiBQCAQ5FKnTh1ER0fj7t27ePbsGc6cOSPXVLV79246Sq6iaRIIhB8dooARCAQCQS5qampwdnbG0aNHERMTg3HjxuHXX3+Fubk5GAwGkpOTceHCBdy/fx+AKE3JhAkTqrnVBELtgChgBAKBQFDIggULkJCQgJCQEMTExEg520vStWtX/P333xVeq5BA+NEhChiBQCAQFKKlpYUDBw7g1q1bCAgIQHR0NDgcDtTU1GBqagoLCwsMHz4c/fr1q3GL2hMINRkSBUkgEAgEAoGgYshSRAQCgUAgEAgqhihgBAKBQCAQCCqGKGAEAoFAIBAIKoYoYAQCgUAgEAgqhihgBAKBQCAQCCqGKGAEAoFAIBAIKoYoYAQCgUAgEAgqhihgBAKBQCAQCCrmfxZLbtPy3TjAAAAAAElFTkSuQmCC"
          },
          "metadata": {}
        }
      ],
      "source": [
        "from sklearn.model_selection import learning_curve\n",
        "\n",
        "# Build learning curve\n",
        "train_sizes, train_scores, test_scores = learning_curve(\n",
        "    model,\n",
        "    X=X_train,\n",
        "    y=y_train,\n",
        "    cv=10,\n",
        "    shuffle=True,\n",
        "    random_state=42,\n",
        "    n_jobs=4,\n",
        "    # We have (0.8*1440)/10 = 115/1150 training samples, 1035/1150 test samples\n",
        "    train_sizes=[10,100,300,500,800, 1000] # bounded at 1035 for this dataset for 10-fold cv\n",
        ")\n",
        "# Get errors for learning curve\n",
        "train_scores_mean = np.mean(train_scores, axis=1)\n",
        "train_scores_std = np.std(train_scores, axis=1)\n",
        "test_scores_mean = np.mean(test_scores, axis=1)\n",
        "test_scores_std = np.std(test_scores, axis=1)\n",
        "\n",
        "# Plot learning curve\n",
        "plt.figure()\n",
        "plt.fill_between(train_sizes, train_scores_mean - train_scores_std,\n",
        "                     train_scores_mean + train_scores_std, alpha=0.1,\n",
        "                     color=\"r\")\n",
        "plt.fill_between(train_sizes, test_scores_mean - test_scores_std,\n",
        "                     test_scores_mean + test_scores_std, alpha=0.1,\n",
        "                     color=\"g\")\n",
        "plt.plot(train_sizes, train_scores_mean, 'o-', color=\"r\",\n",
        "             label=\"Training score\")\n",
        "plt.plot(train_sizes, test_scores_mean, 'o-', color=\"g\",\n",
        "             label=\"Cross-validation score\")\n",
        "plt.title('Learning Curve for MLP Model')\n",
        "plt.xlabel('Number of Training Samples')\n",
        "plt.ylabel('Score')\n",
        "plt.legend(loc=\"best\")\n",
        "plt.show()"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "ma7shatAL9hs"
      },
      "source": [
        "One thing the learning curve tells us is that the size of the dataset isn't our biggest issue - our **accuracy is plateauing and wouldn't seriously benefit from a larger training set** - maybe 5% at most with a gigantic dataset. Again, the gap between the training score curve and cross-validation score curve shows us that the model has an extremely high variance and - it scores perfectly on training data but poorly in cross-validation because it is seriously overfit - the model does not generalize well at all on test data.\n",
        "\n",
        "It has become clear that an MLP network may not be the best choice of model for this task - it seems to not have the complexity we would need to properly discriminate the differences between our features w.r.t. emotion. It appears at this point that we would need a considerably more sophisticated deep neural net to get better performance on this dataset."
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "SyAD_OsSL9hs"
      },
      "source": [
        "## Conclusion\n",
        "\n",
        "The MLPClassifier is powerful in that it achieves appreciable performance with relatively little effort invested in exploratory analysis, hyperparameter optimization, and model architechture; especially taking advantage of grid search, and much more so when we know how to tune each hyperparameter individually.\n",
        "\n",
        "We're going to have to explore more complicated deep learning methods to get real performance on this dataset. **Long-Short-Term-Memory Recurrent Neural Networks (LSTM RNNs) and Convolutional Neural Networks (CNNs) are excellent DNN candidates for audio data classification: LSTM RNNs because of their excellent ability to interpret sequential data such as the audio waveform represented as a time series; CNNs because features engineered on audio data such as spectrograms have marked resemblance to images, in which CNNs excel at recognizing and discriminating between distinct patterns.**\n"
      ]
    }
  ],
  "metadata": {
    "accelerator": "GPU",
    "colab": {
      "provenance": [],
      "include_colab_link": true
    },
    "kernelspec": {
      "display_name": "PyCharm (Speech Classifier)",
      "language": "python",
      "name": "pycharm-6a34225"
    },
    "language_info": {
      "codemirror_mode": {
        "name": "ipython",
        "version": 3
      },
      "file_extension": ".py",
      "mimetype": "text/x-python",
      "name": "python",
      "nbconvert_exporter": "python",
      "pygments_lexer": "ipython3",
      "version": "3.8.3"
    }
  },
  "nbformat": 4,
  "nbformat_minor": 0
}