Model

This class provides functionality for handling all model methods. This class runs the whole pipeline and uses a Preprocessor instance to obtain batches.

With an initialized Neural Network model instance, it is possible to run training, prediction and evaluations.

Methods

Initialization

Neural_Network(preprocessor, architecture=Architecture(),
               loss=tversky_loss, metrics=[dice_soft],
               learninig_rate=0.0001, batch_queue_size=2,
               workers=1, gpu_number=1)

Initialization function for creating a Neural Network (model) object.

Arguments:

• preprocessor: Preprocessor class instance which provides the Neural Network with batches.
• architecture: Instance of a neural network model Architecture class instance. By default, a standard U-Net is used as Architecture.
• loss: The Metric function which is used as loss for training.
• metrics: List of one or multiple Metric Functions, which will be shown during training.
• learninig_rate: Learning rate in which weights of the neural network will be updated.
• batch_queue_size: The batch queue size is the number of previously prepared batches in the cache during runtime.
• workers: Number of workers/threads which preprocess batches during runtime.
• gpu_number: Number of GPUs, which will be used for training.

Returns:
A Neural Network Model class object.

Example:

model = miscnn.Neural_Network(preprocessor=pp)
model.train(sample_list[0:100], epochs=50)

---

train

train(sample_list, epochs=20, iterations=None, callbacks=[])

Fitting function for the Neural Network model using the provided list of sample indices.

Arguments:

• a: asd
• b: asd
• c: asd

Returns:
A Neural Network Model class object.

Example:

function

---

predict

get_indiceslist()

1-2 sentence description

Arguments:

• a: asd
• b: asd
• c: asd

Returns:
A Neural Network Model class object.

Example:

function

---

evaluate

get_indiceslist()

1-2 sentence description

Arguments:

• a: asd
• b: asd
• c: asd

Returns:
A Neural Network Model class object.

Example:

function

---

reset_weights

get_indiceslist()

1-2 sentence description

Arguments:

• a: asd
• b: asd
• c: asd

Returns:
A Neural Network Model class object.

Example:

function

---

dump

get_indiceslist()

1-2 sentence description

Arguments:

• a: asd
• b: asd
• c: asd

Returns:
A Neural Network Model class object.

Example:

function

---

load

get_indiceslist()

1-2 sentence description

Arguments:

• a: asd
• b: asd
• c: asd

Returns:
A Neural Network Model class object.

Example:

function

---