Skip to content
This repository has been archived by the owner on May 31, 2023. It is now read-only.

Latest commit

 

History

History
297 lines (276 loc) · 20.8 KB

REMAIN_SORTED_PAPER.md

File metadata and controls

297 lines (276 loc) · 20.8 KB

Intro

This part save the papers collected from arXiv, paper will be deleted if I judge it is not good or not accepted to the top conference over one year.

No papers

Remain checkout

  • [IJCAI 2021] MEDA: Meta-Learning with Data Augmentation for Few-Shot Text Classification

  • [IJCAI 2021] Cross-Domain Few-Shot Classification via Adversarial Task Augmentation

  • [IJCAI 2021] Learn from Concepts: Towards the Purified Memory for Few-shot Learning

  • [IJCAI 2021] MFNP: A Meta-optimized Model for Few-shot Next POI Recommendation

  • [IJCAI 2021] Few-Shot Learning with Part Discovery and Augmentation from Unlabeled Images

  • [IJCAI 2021] Few-Shot Partial-Label Learning

  • [IJCAI 2021] Conditional Self-Supervised Learning for Few-Shot Classification

  • [ICML 2021] Self-Damaging Contrastive Learning

  • [ICML 2021] Demonstration-Conditioned Reinforcement Learning for Few-Shot Imitation

  • [ICML 2021] Parameterless Transductive Feature Re-representation for Few-Shot Learning

  • [ICML 2021] (paper) Unsupervised Embedding Adaptation via Early-Stage Feature Reconstruction for Few-Shot Classification

  • [ICML 2021] GP-Tree: A Gaussian Process Classifier for Few-Shot Incremental Learning

  • [ICML 2021] Few-Shot Conformal Prediction with Auxiliary Tasks

  • [ICML 2021] Learning a Universal Template for Few-shot Dataset Generalization

  • [ICML 2021] Calibrate Before Use: Improving Few-shot Performance of Language Models

  • [ICML 2021] Few-shot Language Coordination by Modeling Theory of Mind

  • [ICML 2021] A large-scale benchmark for few-shot program induction and synthesis

  • [ICML 2021] Bridging Multi-Task Learning and Meta-Learning: Towards Efficient Training and Effective Adaptation

  • [ICML 2021] Addressing Catastrophic Forgetting in Few-Shot Problems

  • [ICCV 2021] Z-Score Normalization, Hubness, and Few-Shot Learning

  • [ICCV 2021] Learning Meta-Class Memory for Few-Shot Semantic Segmentation

  • [ICCV 2021] Shallow Bayesian Meta Learning for Real-World Few-Shot Recognition

  • [ICCV 2021] Hierarchical Graph Attention Network for Few-Shot Visual-Semantic Learning

  • [ICCV 2021] Query Adaptive Few-Shot Object Detection With Heterogeneous Graph Convolutional Networks

  • [ICCV 2021] Recurrent Mask Refinement for Few-Shot Medical Image Segmentation

  • [ICCV 2021] H3D-Net: Few-Shot High-Fidelity 3D Head Reconstruction

  • [ICCV 2021] Putting NeRF on a Diet: Semantically Consistent Few-Shot View Synthesis

  • [ICCV 2021] Just a Few Points Are All You Need for Multi-View Stereo: A Novel Semi-Supervised Learning Method for Multi-View Stereo

  • [ICCV 2021] Hypercorrelation Squeeze for Few-Shot Segmenation

  • [ICCV 2021] Few-Shot Semantic Segmentation With Cyclic Memory Network

  • [ICCV 2021] Binocular Mutual Learning for Improving Few-Shot Classification

  • [ICCV 2021] Transductive Few-Shot Classification on the Oblique Manifold

  • [ICCV 2021] Task-Aware Part Mining Network for Few-Shot Learning

  • [ICCV 2021] A Multi-Mode Modulator for Multi-Domain Few-Shot Classification

  • [ICCV 2021] LoFGAN: Fusing Local Representations for Few-Shot Image Generation

  • [ICCV 2021] Unsupervised Few-Shot Action Recognition via Action-Appearance Aligned Meta-Adaptation

  • [ICCV 2021] A Hierarchical Transformation-Discriminating Generative Model for Few Shot Anomaly Detection

  • [ICCV 2021] Synthesized Feature Based Few-Shot Class-Incremental Learning on a Mixture of Subspaces

  • [ICCV 2021] Pseudo-Loss Confidence Metric for Semi-Supervised Few-Shot Learning

  • [ICCV 2021] DeFRCN: Decoupled Faster R-CNN for Few-Shot Object Detection

  • [ICCV 2021] Curvature Generation in Curved Spaces for Few-Shot Learning

  • [ICCV 2021] Mining Latent Classes for Few-Shot Segmentation

  • [ICCV 2021] Simpler Is Better: Few-Shot Semantic Segmentation With Classifier Weight Transformer

  • [ICCV 2021] Iterative Label Cleaning for Transductive and Semi-Supervised Few-Shot Learning

  • [ICCV 2021] Variational Feature Disentangling for Fine-Grained Few-Shot Classification

  • [ICCV 2021] Relational Embedding for Few-Shot Classification

  • [ICCV 2021] Generalized and Incremental Few-Shot Learning by Explicit Learning and Calibration Without Forgetting

  • [ICCV 2021] On the Importance of Distractors for Few-Shot Classification

  • [ICCV 2021] Mixture-Based Feature Space Learning for Few-Shot Image Classification

  • [ICCV 2021] Coarsely-Labeled Data for Better Few-Shot Transfer

  • [ICCV 2021] Meta-Baseline: Exploring Simple Meta-Learning for Few-Shot Learning

  • [ICCV 2021] Video Pose Distillation for Few-Shot, Fine-Grained Sports Action Recognition

  • [ICCV 2021] Boosting the Generalization Capability in Cross-Domain Few-Shot Learning via Noise-Enhanced Supervised Autoencoder

  • [ICCV 2021] Meta Navigator: Search for a Good Adaptation Policy for Few-Shot Learning

  • [ICCV 2021] Few-Shot Image Classification: Just Use a Library of Pre-Trained Feature Extractors and a Simple Classifier

  • [ICCV 2021] Few-Shot and Continual Learning With Attentive Independent Mechanisms

  • [ICCV 2021] Meta-Learning With Task-Adaptive Loss Function for Few-Shot Learning

  • [ICCV 2021] Universal Representation Learning From Multiple Domains for Few-Shot Classification

  • [ICCV 2021] Universal-Prototype Enhancing for Few-Shot Object Detection

  • [ICCV 2021] UVStyle-Net: Unsupervised Few-Shot Learning of 3D Style Similarity Measure for B-Reps

  • [ICCV 2021] Improve Unsupervised Pretraining for Few-Label Transfer

  • [ICCV 2021] Partner-Assisted Learning for Few-Shot Image Classification

  • [ICCV 2021] ORBIT: A Real-World Few-Shot Dataset for Teachable Object Recognition

  • [ICCV 2021] Learned Spatial Representations for Few-Shot Talking-Head Synthesis

  • [ICCV 2021] Multiple Heads Are Better Than One: Few-Shot Font Generation With Multiple Localized Experts

  • [ICCV 2021] FLAR: A Unified Prototype Framework for Few-Sample Lifelong Active Recognition

  • [ICCV 2021] Few-Shot Visual Relationship Co-Localization

Remain Read Papers

  • [arXiv 2020] Improving out-of-distribution generalization via multi-task self-supervised pretraining
    • finished
    • no compare to mini-imagenet, so hard to compare
    • exist code

Other Conference

  • [WACV 2020] Charting the Right Manifold: Manifold Mixup for Few-shot Learning
  • [CVPR 2020 Workshop] MA 3 : Model Agnostic Adversarial Augmentation for Few Shot learning
  • [Neuro Computing 2020] Revisiting Metric Learning for Few-Shot Image Classification
  • [IJCNN 2020] RelationNet2: Deep Comparison Columns for Few-Shot Learning
  • [CVPR 2020 Workshop] Meta-Learning for Few-Shot Land Cover Classification
  • [OCEANS 2020] A Comparison of Few-Shot Learning Methods for Underwater Optical and Sonar Image Classification
    • K-means enhance the prototypes, similar to my previous papers.
  • [InterSpeech 2020] AdaDurIAN: Few-shot Adaptation for Neural Text-to-Speech with DurIAN
  • [ICML 2020 Workshop]Covariate Distribution Aware Meta-learning
  • [ESANN 2020] Zero-shot and few-shot time series forecasting with ordinal regression recurrent neural networks
  • [TIP 2020] BSNet: Bi-Similarity Network for Few-shot Fine-grained Image Classification
  • [ICASSP 2021] Domain Adaptation for Learning Generator from Paired Few-Shot Data
  • [ICEMS 2021] Few-Shot Bearing Anomaly Detection Based on Model-Agnostic Meta-Learning
  • [ICASSP2021] Few-shot Image Classification with Multi-Facet Prototypes

Arxiv

Summary

  • [arXiv 2020] A Concise Review of Recent Few-shot Meta-learning Methods
    • Change the Methods into four methods. (Basically exclude metric-based such as ProtoNet)
      • Learning an Initialization
      • Generation of Parameters
      • Learning an Optimizer (doubt for this, maybe sort to the second)
      • Memory-based Methods

Image Classification

  • [arXiv 2020] (paper code) Exploiting Unsupervised Inputs for Accurate Few-Shot Classification

    • 85% Graph
    • wait author to refine the paper
  • [arXiv 2020] (paper) Prior-Knowledge and Attention based Meta-Learning for Few-Shot Learning

    • add Very Very simple attention(almost like SENet's attention model)
    • add addition model to assist
    • 1 per improve on PN
  • [arXiv 2020] (paper) Few-Shot Few-Shot Learning and the role of Spatial Attention

    • reported 80% acc
    • Interesting
  • [arXiv 2020] Few-Shot Learning with Geometric Constraints

    • main contribution is on remain accuracy for both novel and base
    • outperform in both situations on miniImagenet
  • [arXiv 2020] A New Meta-Baseline for Few-Shot Learning

  • [arXiv 2020] AdarGCN: Adaptive Aggregation GCN for Few-Shot Learning

    • a new circumstance with noise input
  • [arXiv 2020] Prototype Rectification for Few-Shot Learning

    • Using query set to enhance prototype, good results in 1-shot 70%, however, potential model leaky problem, wait for opensource
  • [arXiv 2020] [exist code] Transductive Few-shot Learning with Meta-Learned Confidence

    • 78 1-shot 86 5-shot
  • [arXiv 2020] Self-Augmentation: Generalizing Deep Networks to Unseen Classes for Few-Shot Learning

    • 65 1-shot 82 5-shot
    • deep metric learning & cutmix
  • [arXiv 2020] Unsupervised Few-shot Learning via Distribution Shift-based Augmentation

  • [arXiv 2020] Meta-Meta-Classification for One-Shot Learning

  • [arXiv 2020] Divergent Search for Few-Shot Image Classification

  • [arXiv 2020] Physarum Powered Differentiable Linear Programming Layers and Applications

    • An plug and play layer, FC-100 improve Cifar-100 FS for 1% on MetaOptSVM
  • [arXiv 2020] Generalized Reinforcement Meta Learning for Few-Shot Optimization

    • This paper's motivation like "Empirical Bayes Transductive Meta-Learning with Synthetic Gradients in ICLR 2020, both of them use a mechanism to estimate or synthesis the gradients, so if you interesting in this paper, you'd better have look it that one.
    • 71% with Resnet on mini-Imagenet (not so impressive)
  • [arXiv 2020] Bayesian Online Meta-Learning with Laplace Approximation

    • continue learning
  • [arXiv 2020] ONE OF THESE (FEW) THINGS IS NOT LIKE THE OTHERS

    • image classification task, however need to classify the outline, like cross-domain settings
    • they define a "junk" classes specific, may not significant
  • [arXiv 2020] Compositional Few-Shot Recognition with Primitive Discovery and Enhancing

    • 63.21 ± 0.78% for 1-shot on mini-ImageNet
    • interest, they improve the results by decompose the image into several compositions, sound reasonable
  • [arXiv 2020] Looking back to lower-level information in few-shot learning

    • arguing that output of each layer in the backbone's could use to learn
    • build TPN on each layer of the graph
    • Improve TPN for 1~2% for mini-Imagenet and tiered-Imagenet compare to TPN
  • [arXiv 2020] TOAN: Target-Oriented Alignment Network for Fine-Grained Image Categorization with Few Labeled Samples

    • very similar to one of work in CVPR 2019 -> Revisiting Local Descriptor based Image-to-Class Measure for Few-shot Learning
    • Not compare in mini-ImageNet
    • same to "Low-Rank Pairwise Alignment Bilinear Network For Few-Shot Fine-Grained Image Classification"
  • [arXiv 2020] Few-shot Learning for Domain-specific Fine-grained Image Classification

  • [arXiv 2020] Distributionally Robust $k$-Nearest Neighbors for Few-Shot Learning

    • Seem mathmatically methods, will verify it later
  • [arXiv 2020] Learning to Learn Kernels with Variational Random Features

    • LSTM to adjust kernel
    • 54% for 1-shot 67.8% for 5-shot
  • [arXiv 2020] Prototype Rectification for Few-Shot Learning

    • 70.31% fot 1-shot and 81.89% for 5-shot
  • [arXiv 2020] Convergence of Meta-Learning with Task-Specific Adaptation over Partial Parameters

    • update paritial parameters in inner loop
    • improve for MAML
  • [arXiv 2020] (code)Self-supervised Knowledge Distillation for Few-shot Learning

    • learn feature into stage, pretrain + self-traning
    • 67% for 1-shot on mini-Imagenet with ResNet12
  • [arXiv 2020] Unsupervised Meta-Learning through Latent-Space Interpolation in Generative Models

    • unsupervised, compare to DeepCluster
  • [arXiv 2020] (code) Leveraging the Feature Distribution in Transfer-based Few-Shot Learning

    • 82% for 1-shot 88% for 5-shot WRN
    • First force the encoded feature to satisfy a certain distribution, then use specific algorithm designed for the distribution
  • [arXiv 2020] Graph Meta Learning via Local Subgraphs

    • target Graph query the related graphs
  • [arXiv 2020] Adaptive-Step Graph Meta-Learner for Few-Shot Graph Classification

    • Graph level query
  • [arXiv 2020] Improving Few-Shot Visual Classification with Unlabelled Examples

    • Cluster-based proto finetune methods
    • 80% for 1-shot on mini-Imagenet
  • [arXiv 2020] Improving Few-Shot Learning using Composite Rotation based Auxiliary Task

    • Result is impressive
    • 68% for 1-shot 84 for 5-shot for mini-Imagenet (ResNet 18)
    • rotation image to perform self-supervise learning
  • [arXiv 2020] (code) A Universal Representation Transformer Layer for Few-Shot Image Classification

    • The author have one paper accepted in TKDE
    • Idea is interesting
    • Universal Representation Transformer (URT) layer, that meta-learns to leverage universal features for few-shot classification by dynamically re-weighting and composing the most appropriate domain-specific representations
  • [arXiv 2020] (code) Transductive Information Maximization For Few-Shot Learning

    • Our method maximizes the mutual information between the query features and predictions of a few-shot task, subject to supervision constraints from the support set.
    • Results very high in WRN-28 77.8% and 87% for 1-shot and 5-shot on mini-ImageNet

Generation

  • [arXiv 2020] MatchingGAN: Matching-based Few-shot Image Generation

Object Detection & Tracking

  • [arXiv 2020] Meta R-CNN : Towards General Solver for Instance-level Few-shot Learning
  • [arXiv 2020] Weakly-supervised Any-shot Object Detection
  • [arXiv 2020] MOTS: Multiple Object Tracking for General Categories Based On Few-Shot Method
    • nearly same to prototype networks
  • [arXiv 2020] Few-shot Object Detection on Remote Sensing Images

Segmentation

  • [arXiv 2020] On the Texture Bias for Few-Shot CNN Segmentation
  • [arXiv 2020] [exist code] Learning to Segment the Tail
  • [arXiv 2020] Semi-supervised few-shot learning for medical image segmentation
  • [arXiv 2020] Objectness-Aware One-Shot Semantic Segmentation
  • [arXiv 2020] Self-Supervised Tuning for Few-Shot Segmentation
  • [arXiv 2020] Prototype Refinement Network for Few-Shot Segmentation
  • [arXiv 2020] Few-Shot Semantic Segmentation Augmented with Image-Level Weak Annotations

NLP

  • [arXiv 2020] Few-shot Natural Language Generation for Task-Oriented Dialog

  • [arXiv 2020] SOLOIST: Few-shot Task-Oriented Dialog with A Single Pre-trained Auto-regressive Model

  • [arXiv 2020] Prototypical Q Networks for Automatic Conversational Diagnosis and Few-Shot New Disease Adaption

    • Using a dialog to simulate patient and doctors' conversation to finally give a diagnosis
  • [arXiv 2020] BOFFIN TTS: FEW-SHOT SPEAKER ADAPTATION BY BAYESIAN OPTIMIZATION

    • Text to Speech
  • [arXiv 2020] MICK: A Meta-Learning Framework for Few-shot Relation Classification with Little Training Data

    • Relation Classification
  • [arXiv 2020] Logic2Text: High-Fidelity Natural Language Generation from Logical Forms

  • [arXiv 2020] Learning to Few-Shot Learn Across Diverse Natural Language Classification Tasks

    • LEO 风格的文本分类
    • Structure like LEO, named LEOPARD
  • [arXiv 2020] Meta Fine-Tuning Neural Language Models for Multi-Domain Text Mining

  • [arXiv 2020] Cross-lingual Zero- and Few-shot Hate Speech Detection Utilising Frozen Transformer Language Models and AXEL

  • [arXiv 2020] Few-Shot Natural Language Generation by Rewriting Templates

  • [arXiv 2020] Pay Attention to What You Read: Non-recurrent Handwritten Text-Line Recognition

  • [arXiv 2020] CG-BERT: Conditional Text Generation with BERT for Generalized Few-shot Intent Detection

Cross-Domain

  • [arXiv 2020] Towards Fair Cross-Domain Adaptation via Generative Learning
  • [arXiv 2020] Cross-Domain Few-Shot Learning with Meta Fine-Tuning
  • [arXiv 2020] Feature Transformation Ensemble Model with Batch Spectral Regularization for Cross-Domain Few-Shot Classification
  • [arXiv 2020] Cross-Domain Few-Shot Learning with Meta Fine-Tuning
  • [arXiv 2020] Large Margin Mechanism and Pseudo Query Set on Cross-Domain Few-Shot Learning
  • [arXiv 2020] M3P: Learning Universal Representations via Multitask Multilingual Multimodal Pre-training

Uncertainty

  • [arXiv 2020] Calibrated neighborhood aware confidence measure for deep metric learning
    • The approach approximates the distribution of data points for each class using a Gaussian kernel smoothing function.
    • They operate the uncertainty measure methods into three branches Calibration on the held-out validation data, Bayesian approximation / Support set based uncertainty estimation
    • The gt used for uncertainty is measured by euclidean distances and I doublet the uncertainty measured by euclidean in high dimensional spaces is accurate.

Application

  • [arXiv 2020] DAWSON: A Domain Adaptive Few Shot Generation Framework

    • generate music, a project under cs236 in Stanford university
  • [arXiv 2020] Few-Shot Scene Adaptive Crowd Counting Using Meta-Learning

  • [arXiv 2020] Meta-Learning Initializations for Low-Resource Drug Discovery

  • [arXiv 2020] An Open-set Recognition and Few-Shot Learning Dataset for Audio Event Classification in Domestic Environments

  • [arXiv 2020] Additive Angular Margin for Few Shot Learning to Classify Clinical Endoscopy Images

  • [arXiv 2020] Domain-Adaptive Few-Shot Learning

  • [arXiv 2020] Efficient Intent Detection with Dual Sentence Encoders

  • [arXiv 2020] Zero-Shot Cross-Lingual Transfer with Meta Learning

  • [arXiv 2020] From Zero to Hero: On the Limitations of Zero-Shot Cross-Lingual Transfer with Multilingual Transformers

  • [arXiv 2020] An Open-set Recognition and Few-Shot Learning Dataset for Audio Event Classification in Domestic Environments

  • [arXiv 2020 wip] PAC-BAYESIAN META-LEARNING WITH IMPLICIT PRIOR

    • 63 for 1shot, 78 for 5-shot
    • LEO branch
  • [arXiv 2020] Revisiting Few-shot Activity Detection with Class Similarity Control

  • [arXiv 2020] Meta-Learning for Few-Shot NMT Adaptation

  • [arXiv 2020] SSHFD: Single Shot Human Fall Detection with Occluded Joints Resilience

  • [arXiv 2020] Gradient-based Data Augmentation for Semi-Supervised Learning

  • [arXiv 2020] Alleviating the Incompatibility between Cross Entropy Loss and Episode Training for Few-shot Skin Disease Classification

  • [arXIv 2020] TAEN: Temporal Aware Embedding Network for Few-Shot Action Recognition

  • [arXiv 2020] Signal Level Deep Metric Learning for Multimodal One-Shot Action Recognition

  • [arXiv 2020] ST2: Small-data Text Style Transfer via Multi-task Meta-Learning

  • [arXiv 2020] Learning to Classify Intents and Slot Labels Given a Handful of Examples

  • [arXiv 2020] PuzzLing Machines: A Challenge on Learning From Small Data

  • [arXiv 2020] Learning to Learn to Disambiguate: Meta-Learning for Few-Shot Word Sense Disambiguation

  • [arXiv 2020] Few-Shot Learning for Abstractive Multi-Document Opinion Summarization

  • [arXiv 2020] Interactive Video Stylization Using Few-Shot Patch-Based Training

  • [arXiv 2020] MAD-X: An Adapter-based Framework for Multi-task Cross-lingual Transfer

  • [arXiv 2020] Self-Training with Improved Regularization for Few-Shot Chest X-Ray Classification

  • [arXiv 2020] Combining Deep Learning with Geometric Features for Image based Localization in the Gastrointestinal Tract

  • [arXiv 2020] SSM-Net for Plants Disease Identification in LowData Regime

    • disease in agricultural
  • [arXiv 2020] Extensively Matching for Few-shot Learning Event Detection

  • [arXiv 2020] Text Recognition in Real Scenarios with a Few Labeled Samples

    • They trying to address the text retrieval problems when Targe domain is nosey
  • [arXiv 2020] Inductive Relational Matrix Completion

    • Recommend System cold-start relevant
  • Learning to Profile: User Meta-Profile Network for Few-Shot Learning

    • Few-shot in recommend system, basically focus on encoding more efficient.
    • Pretty Interesting , may read later