【增量学习】很棒的增量学习/终身学习

Survey

• Replay in Deep Learning: Current Approaches and Missing Biological Elements (Neural Computation 2021) [paper]
• Online Continual Learning in Image Classification: An Empirical Survey (Neurocomputing 2021) [paper] [code]
• Continual Lifelong Learning in Natural Language Processing: A Survey (COLING 2020) [paper]
• Class-incremental learning: survey and performance evaluation (arXiv 2020) [paper] [code]
• A Comprehensive Study of Class Incremental Learning Algorithms for Visual Tasks (Neural Networks) [paper] [code]
• A continual learning survey: Defying forgetting in classification tasks (TPAMI 2021) [paper] [arxiv]
• Continual Lifelong Learning with Neural Networks: A Review (Neural Networks) [paper]
• Three scenarios for continual learning (arXiv 2019) [paper][code]

Papers

2022

• Learning to Imagine: Diversify Memory for Incremental Learning using Unlabeled Data (CVPR2022) [paper]
• General Incremental Learning with Domain-aware Categorical Representations (CVPR2022) [paper]
• Constrained Few-shot Class-incremental Learning (CVPR2022) [paper]
• Overcoming Catastrophic Forgetting in Incremental Object Detectionvia Elastic Response Distillation (CVPR2022) [paper]
• Class-Incremental Learning with Strong Pre-trained Models (CVPR2022) [paper]
• Energy-based Latent Aligner for Incremental Learning (CVPR2022) [paper] [code]
• Meta-attention for ViT-backed Continual Learning (CVPR2022) [paper] [code]
• Learning to Prompt for Continual Learning (CVPR2022) [paper] [code]
• On Generalizing Beyond Domains in Cross-Domain Continual Learning (CVPR2022) [paper]
• Probing Representation Forgetting in Supervised and Unsupervised Continual Learning (CVPR2022) [paper]
• Incremental Transformer Structure Enhanced Image Inpainting with Masking Positional Encoding (CVPR2022) [paper] [code]
• Mimicking the Oracle: An Initial Phase Decorrelation Approach for Class Incremental Learning (CVPR2022) [paper] [code]
• Forward Compatible Few-Shot Class-Incremental Learning (CVPR2022) [paper] [code]
• Self-Sustaining Representation Expansion for Non-Exemplar Class-Incremental Learning (CVPR2022) [paper]
• DyTox: Transformers for Continual Learning with DYnamic TOken eXpansion (CVPR2022) [paper]
• Federated Class-Incremental Learning (CVPR2022) [paper] [code]
• Representation Compensation Networks for Continual Semantic Segmentation (CVPR2022) [paper]
• Self-training for class-incremental semantic segmentation (TNNLS2022) [paper]
• Continual Sequence Generation with Adaptive Compositional Modules (ACL2022) [paper]
• Learngene: From Open-World to Your Learning Task (AAAI2022) [paper] [code]
• MgSvF: Multi-Grained Slow vs. Fast Framework for Few-Shot Class-Incremental Learning (TPAMI2022) [paper]
• Rethinking the Representational Continuity: Towards Unsupervised Continual Learning (ICLR2022) [paper]
• Continual Learning with Filter Atom Swapping (ICLR2022) [paper]
• Continual Learning with Recursive Gradient Optimization (ICLR2022) [paper]
• TRGP: Trust Region Gradient Projection for Continual Learning (ICLR2022) [paper]
• Looking Back on Learned Experiences For Class/task Incremental Learning (ICLR2022) [paper]
• Continual Normalization: Rethinking Batch Normalization for Online Continual Learning (ICLR2022) [paper]
• Model Zoo: A Growing Brain That Learns Continually (ICLR2022) [paper]
• Learning curves for continual learning in neural networks: Self-knowledge transfer and forgetting (ICLR2022) [paper]
• Memory Replay with Data Compression for Continual Learning (ICLR2022) [paper]
• Learning Fast, Learning Slow: A General Continual Learning Method based on Complementary Learning System (ICLR2022) [paper]
• Online Coreset Selection for Rehearsal-based Continual Learning (ICLR2022) [paper]
• Pretrained Language Model in Continual Learning: A Comparative Study (ICLR2022) [paper]
• Online Continual Learning on Class Incremental Blurry Task Configuration with Anytime Inference (ICLR2022) [paper]
• New Insights on Reducing Abrupt Representation Change in Online Continual Learning (ICLR2022) [paper]
• Towards Continual Knowledge Learning of Language Models (ICLR2022) [paper]
• CLEVA-Compass: A Continual Learning Evaluation Assessment Compass to Promote Research Transparency and Comparability (ICLR2022) [paper]
• CoMPS: Continual Meta Policy Search (ICLR2022) [paper]
• Information-theoretic Online Memory Selection for Continual Learning (ICLR2022) [paper]
• Subspace Regularizers for Few-Shot Class Incremental Learning (ICLR2022) [paper]
• LFPT5: A Unified Framework for Lifelong Few-shot Language Learning Based on Prompt Tuning of T5 (ICLR2022) [paper]
• Effect of scale on catastrophic forgetting in neural networks (ICLR2022) [paper]
• Dataset Knowledge Transfer for Class-Incremental Learning without Memory (WACV2022) [paper]
• Knowledge Capture and Replay for Continual Learning (WACV2022) [paper]
• Online Continual Learning via Candidates Voting (WACV2022) [paper]

2021

• Incremental Object Detection via Meta-Learning (TPAMI 2021) [paper] [code]
• Memory efficient class-incremental learning for image classification (TNNLS 2021) [paper]
• Class-Incremental Learning via Dual Augmentation (NeurIPS2021) [paper]
• SSUL: Semantic Segmentation with Unknown Label for Exemplar-based Class-Incremental Learning (NeurIPS2021) [paper]
• RMM: Reinforced Memory Management for Class-Incremental Learning (NeurIPS2021) [paper]
• Overcoming Catastrophic Forgetting in Incremental Few-Shot Learning by Finding Flat Minima (NeurIPS2021) [paper]
• Lifelong Domain Adaptation via Consolidated Internal Distribution (NeurIPS2021) [paper]
• AFEC: Active Forgetting of Negative Transfer in Continual Learning (NeurIPS2021) [paper]
• Natural continual learning: success is a journey, not (just) a destination (NeurIPS2021) [paper]
• Gradient-based Editing of Memory Examples for Online Task-free Continual Learning (NeurIPS2021) [paper]
• Optimizing Reusable Knowledge for Continual Learning via Metalearning (NeurIPS2021) [paper]
• Formalizing the Generalization-Forgetting Trade-off in Continual Learning (NeurIPS2021) [paper]
• Learning where to learn: Gradient sparsity in meta and continual learning (NeurIPS2021) [paper]
• Flattening Sharpness for Dynamic Gradient Projection Memory Benefits Continual Learning (NeurIPS2021) [paper]
• Posterior Meta-Replay for Continual Learning (NeurIPS2021) [paper]
• Continual Auxiliary Task Learning (NeurIPS2021) [paper]
• Mitigating Forgetting in Online Continual Learning with Neuron Calibration (NeurIPS2021) [paper]
• BNS: Building Network Structures Dynamically for Continual Learning (NeurIPS2021) [paper]
• DualNet: Continual Learning, Fast and Slow (NeurIPS2021) [paper]
• BooVAE: Boosting Approach for Continual Learning of VAE (NeurIPS2021) [paper]
• Generative vs. Discriminative: Rethinking The Meta-Continual Learning (NeurIPS2021) [paper]
• Achieving Forgetting Prevention and Knowledge Transfer in Continual Learning (NeurIPS2021) [paper]
• Bridging Non Co-occurrence with Unlabeled In-the-wild Data for Incremental Object Detection (NeurIPS, 2021) [paper] [code]
• SS-IL: Separated Softmax for Incremental Learning (ICCV, 2021) [paper]
• Striking a Balance between Stability and Plasticity for Class-Incremental Learning (ICCV, 2021) [paper]
• Synthesized Feature based Few-Shot Class-Incremental Learning on a Mixture of Subspaces (ICCV, 2021) [paper]
• Class-Incremental Learning for Action Recognition in Videos (ICCV, 2021) [paper]
• Continual Prototype Evolution:Learning Online from Non-Stationary Data Streams (ICCV, 2021) [paper]
• Rehearsal Revealed: The Limits and Merits of Revisiting Samples in Continual Learning (ICCV, 2021) [paper]
• Co2L: Contrastive Continual Learning (ICCV, 2021) [paper]
• Wanderlust: Online Continual Object Detection in the Real World (ICCV, 2021) [paper]
• Continual Learning on Noisy Data Streams via Self-Purified Replay (ICCV, 2021) [paper]
• Else-Net: Elastic Semantic Network for Continual Action Recognition from Skeleton Data (ICCV, 2021) [paper]
• Detection and Continual Learning of Novel Face Presentation Attacks (ICCV, 2021) [paper]
• Online Continual Learning with Natural Distribution Shifts: An Empirical Study with Visual Data (ICCV, 2021) [paper]
• Continual Learning for Image-Based Camera Localization (ICCV, 2021) [paper]
• Generalized and Incremental Few-Shot Learning by Explicit Learning and Calibration without Forgetting (ICCV, 2021) [paper]
• Always Be Dreaming: A New Approach for Data-Free Class-Incremental Learning (ICCV, 2021) [paper]
• RECALL: Replay-based Continual Learning in Semantic Segmentation (ICCV, 2021) [paper]
• Few-Shot and Continual Learning with Attentive Independent Mechanisms (ICCV, 2021) [paper]
• Learning with Selective Forgetting (IJCAI, 2021) [paper]
• Continuous Coordination As a Realistic Scenario for Lifelong Learning (ICML, 2021) [paper]
• Kernel Continual Learning (ICML, 2021) [paper]
• Variational Auto-Regressive Gaussian Processes for Continual Learning (ICML, 2021) [paper]
• Bayesian Structural Adaptation for Continual Learning (ICML, 2021) [paper]
• Continual Learning in the Teacher-Student Setup: Impact of Task Similarity (ICML, 2021) [paper]
• Continuous Coordination As a Realistic Scenario for Lifelong Learning (ICML, 2021) [paper]
• Federated Continual Learning with Weighted Inter-client Transfer (ICML, 2021) [paper]
• Adapting BERT for Continual Learning of a Sequence of Aspect Sentiment Classification Tasks (NAACL, 2021) [paper]
• Continual Learning for Text Classification with Information Disentanglement Based Regularization (NAACL, 2021) [paper]
• CLASSIC: Continual and Contrastive Learning of Aspect Sentiment Classification Tasks (EMNLP, 2021) [paper][code]
• Co-Transport for Class-Incremental Learning (ACM MM, 2021) [paper]
• Towards Open World Object Detection (CVPR, 2021) [paper] [code] [video]
• Prototype Augmentation and Self-Supervision for Incremental Learning (CVPR, 2021) [paper] [code]
• ORDisCo: Effective and Efficient Usage of Incremental Unlabeled Data for Semi-supervised Continual Learning (CVPR, 2021) [paper]
• Incremental Learning via Rate Reduction (CVPR, 2021) [paper]
• IIRC: Incremental Implicitly-Refined Classification (CVPR, 2021) [paper]
• Continual Adaptation of Visual Representations via Domain Randomization and Meta-learning (CVPR, 2021) [paper]
• Image De-raining via Continual Learning (CVPR, 2021) [paper]
• Continual Learning via Bit-Level Information Preserving (CVPR, 2021) [paper]
• Hyper-LifelongGAN: Scalable Lifelong Learning for Image Conditioned Generation (CVPR, 2021) [paper]
• Lifelong Person Re-Identification via Adaptive Knowledge Accumulation (CVPR, 2021) [paper]
• Distilling Causal Effect of Data in Class-Incremental Learning (CVPR, 2021) [paper]
• Self-Promoted Prototype Refinement for Few-Shot Class-Incremental Learning (CVPR, 2021) [paper]
• Layerwise Optimization by Gradient Decomposition for Continual Learning (CVPR, 2021) [paper]
• Adaptive Aggregation Networks for Class-Incremental Learning (CVPR, 2021) [paper]
• Incremental Few-Shot Instance Segmentation (CVPR, 2021) [paper]
• Efficient Feature Transformations for Discriminative and Generative Continual Learning (CVPR, 2021) [paper]
• On Learning the Geodesic Path for Incremental Learning (CVPR, 2021) [paper]
• Few-Shot Incremental Learning with Continually Evolved Classifiers (CVPR, 2021) [paper]
• Rectification-based Knowledge Retention for Continual Learning (CVPR, 2021) [paper]
• DER: Dynamically Expandable Representation for Class Incremental Learning (CVPR, 2021) [paper]
• Rainbow Memory: Continual Learning with a Memory of Diverse Samples (CVPR, 2021) [paper]
• Training Networks in Null Space of Feature Covariance for Continual Learning (CVPR, 2021) [paper]
• Semantic-aware Knowledge Distillation for Few-Shot Class-Incremental Learning (CVPR, 2021) [paper]
• PLOP: Learning without Forgetting for Continual Semantic Segmentation (CVPR, 2021) [paper]
• Continual Semantic Segmentation via Repulsion-Attraction of Sparse and Disentangled Latent Representations (CVPR, 2021) [paper]
• Online Class-Incremental Continual Learning with Adversarial Shapley Value(AAAI, 2021) [paper] [code]
• Lifelong and Continual Learning Dialogue Systems: Learning during Conversation(AAAI, 2021) [paper]
• Continual learning for named entity recognition(AAAI, 2021) [paper]
• Using Hindsight to Anchor Past Knowledge in Continual Learning(AAAI, 2021) [paper]
• Split-and-Bridge: Adaptable Class Incremental Learning within a Single Neural Network(AAAI, 2021) [paper] [code]
• Curriculum-Meta Learning for Order-Robust Continual Relation Extraction(AAAI, 2021) [paper]
• Continual Learning by Using Information of Each Class Holistically(AAAI, 2021) [paper]
• Gradient Regularized Contrastive Learning for Continual Domain Adaptation(AAAI, 2021) [paper]
• Unsupervised Model Adaptation for Continual Semantic Segmentation(AAAI, 2021) [paper]
• A Continual Learning Framework for Uncertainty-Aware Interactive Image Segmentation(AAAI, 2021) [paper]
• Do Not Forget to Attend to Uncertainty While Mitigating Catastrophic Forgetting(WACV, 2021) [paper]

2020

• Rethinking Experience Replay: a Bag of Tricks for Continual Learning(ICPR, 2020) [paper] [code]
• Continual Learning for Natural Language Generation in Task-oriented Dialog Systems(EMNLP, 2020) [paper]
• Distill and Replay for Continual Language Learning(COLING, 2020) [paper]
• Continual Learning of a Mixed Sequence of Similar and Dissimilar Tasks (NeurIPS2020) [paper] [code]
• Meta-Consolidation for Continual Learning (NeurIPS2020) [paper]
• Understanding the Role of Training Regimes in Continual Learning (NeurIPS2020) [paper]
• Continual Learning with Node-Importance based Adaptive Group Sparse Regularization (NeurIPS2020) [paper]
• Online Fast Adaptation and Knowledge Accumulation (OSAKA): a New Approach to Continual Learning (NeurIPS2020) [paper]
• Coresets via Bilevel Optimization for Continual Learning and Streaming (NeurIPS2020) [paper]
• RATT: Recurrent Attention to Transient Tasks for Continual Image Captioning (NeurIPS2020) [paper]
• Continual Deep Learning by Functional Regularisation of Memorable Past (NeurIPS2020) [paper]
• Dark Experience for General Continual Learning: a Strong, Simple Baseline (NeurIPS2020) [paper] [code]
• GAN Memory with No Forgetting (NeurIPS2020) [paper]
• Calibrating CNNs for Lifelong Learning (NeurIPS2020) [paper]
• Mitigating Forgetting in Online Continual Learning via Instance-Aware Parameterization (NeurIPS2020) [paper]
• ADER: Adaptively Distilled Exemplar Replay Towards Continual Learning for Session-based Recommendation(RecSys, 2020) [paper]
• Initial Classifier Weights Replay for Memoryless Class Incremental Learning (BMVC2020) [paper]
• Adversarial Continual Learning (ECCV2020) [paper] [code]
• REMIND Your Neural Network to Prevent Catastrophic Forgetting (ECCV2020) [paper] [code]
• Incremental Meta-Learning via Indirect Discriminant Alignment (ECCV2020) [paper]
• Memory-Efficient Incremental Learning Through Feature Adaptation (ECCV2020) [paper]
• PODNet: Pooled Outputs Distillation for Small-Tasks Incremental Learning (ECCV2020) [paper] [code]
• Reparameterizing Convolutions for Incremental Multi-Task Learning Without Task Interference (ECCV2020) [paper]
• Learning latent representions across multiple data domains using Lifelong VAEGAN (ECCV2020) [paper]
• Online Continual Learning under Extreme Memory Constraints (ECCV2020) [paper]
• Class-Incremental Domain Adaptation (ECCV2020) [paper]
• More Classifiers, Less Forgetting: A Generic Multi-classifier Paradigm for Incremental Learning (ECCV2020) [paper]
• Piggyback GAN: Efficient Lifelong Learning for Image Conditioned Generation (ECCV2020) [paper]
• GDumb: A Simple Approach that Questions Our Progress in Continual Learning (ECCV2020) [paper]
• Imbalanced Continual Learning with Partitioning Reservoir Sampling (ECCV2020) [paper]
• Topology-Preserving Class-Incremental Learning (ECCV2020) [paper]
• GraphSAIL: Graph Structure Aware Incremental Learning for Recommender Systems (CIKM2020) [paper]
• OvA-INN: Continual Learning with Invertible Neural Networks (IJCNN2020) [paper]
• XtarNet: Learning to Extract Task-Adaptive Representation for Incremental Few-Shot Learning (ICLM2020) [paper]
• Optimal Continual Learning has Perfect Memory and is NP-HARD (ICML2020) [paper]
• Neural Topic Modeling with Continual Lifelong Learning (ICML2020) [paper]
• Continual Learning with Knowledge Transfer for Sentiment Classification (ECML-PKDD2020) [paper] [code]
• Semantic Drift Compensation for Class-Incremental Learning (CVPR2020) [paper] [code]
• Few-Shot Class-Incremental Learning (CVPR2020) [paper]
• Modeling the Background for Incremental Learning in Semantic Segmentation (CVPR2020) [paper]
• Incremental Few-Shot Object Detection (CVPR2020) [paper]
• Incremental Learning In Online Scenario (CVPR2020) [paper]
• Maintaining Discrimination and Fairness in Class Incremental Learning (CVPR2020) [paper]
• Conditional Channel Gated Networks for Task-Aware Continual Learning (CVPR2020) [paper]
• Continual Learning with Extended Kronecker-factored Approximate Curvature (CVPR2020) [paper]
• iTAML : An Incremental Task-Agnostic Meta-learning Approach (CVPR2020) [paper] [code]
• Mnemonics Training: Multi-Class Incremental Learning without Forgetting (CVPR2020) [paper] [code]
• ScaIL: Classifier Weights Scaling for Class Incremental Learning (WACV2020) [paper]
• Accepted papers(ICLR2020) [paper]
• Brain-inspired replay for continual learning with artificial neural networks (Natrue Communications 2020) [paper] [code]
• Learning to Continually Learn (ECAI 2020) [paper] [code]

2019

• Compacting, Picking and Growing for Unforgetting Continual Learning (NeurIPS2019)[paper][code]
• Increasingly Packing Multiple Facial-Informatics Modules in A Unified Deep-Learning Model via Lifelong Learning (ICMR2019) [paper][code]
• Towards Training Recurrent Neural Networks for Lifelong Learning (Neural Computation 2019) [paper]
• Complementary Learning for Overcoming Catastrophic Forgetting Using Experience Replay (IJCAI2019) [paper]
• IL2M: Class Incremental Learning With Dual Memory (ICCV2019) [paper]
• Incremental Learning Using Conditional Adversarial Networks (ICCV2019) [paper]
• Adaptive Deep Models for Incremental Learning: Considering Capacity Scalability and Sustainability (KDD2019) [paper]
• Random Path Selection for Incremental Learning (NeurIPS2019) [paper]
• Online Continual Learning with Maximal Interfered Retrieval (NeurIPS2019) [paper]
• Meta-Learning Representations for Continual Learning (NeurIPS2019) [paper] [code]
• Overcoming Catastrophic Forgetting with Unlabeled Data in the Wild (ICCV2019) [paper]
• Continual Learning by Asymmetric Loss Approximation with Single-Side Overestimation (ICCV2019) [paper]
• Lifelong GAN: Continual Learning for Conditional Image Generation (ICCV2019) [paper]
• Continual learning of context-dependent processing in neural networks (Nature Machine Intelligence 2019) [paper] [code]
• Large Scale Incremental Learning (CVPR2019) [paper] [code]
• Learning a Unified Classifier Incrementally via Rebalancing (CVPR2019) [paper] [code]
• Learning Without Memorizing (CVPR2019) [paper]
• Learning to Remember: A Synaptic Plasticity Driven Framework for Continual Learning (CVPR2019) [paper]
• Task-Free Continual Learning (CVPR2019) [paper]
• Learn to Grow: A Continual Structure Learning Framework for Overcoming Catastrophic Forgetting (ICML2019) [paper]
• Efficient Lifelong Learning with A-GEM (ICLR2019) [paper] [code]
• Learning to Learn without Forgetting By Maximizing Transfer and Minimizing Interference (ICLR2019) [paper] [code]
• Overcoming Catastrophic Forgetting via Model Adaptation (ICLR2019) [paper]
• A comprehensive, application-oriented study of catastrophic forgetting in DNNs (ICLR2019) [paper]

2018

• Memory Replay GANs: learning to generate images from new categories without forgetting (NIPS2018) [paper] [code]
• Reinforced Continual Learning (NIPS2018) [paper] [code]
• Online Structured Laplace Approximations for Overcoming Catastrophic Forgetting (NIPS2018) [paper]
• Rotate your Networks: Better Weight Consolidation and Less Catastrophic Forgetting (R-EWC) (ICPR2018) [paper] [code]
• Exemplar-Supported Generative Reproduction for Class Incremental Learning (BMVC2018) [paper] [code]
• End-to-End Incremental Learning (ECCV2018) [paper][code]
• Riemannian Walk for Incremental Learning: Understanding Forgetting and Intransigence (ECCV2018)[paper]
• Piggyback: Adapting a Single Network to Multiple Tasks by Learning to Mask Weights (ECCV2018) [paper] [code]
• Memory Aware Synapses: Learning what (not) to forget (ECCV2018) [paper] [code]
• Lifelong Learning via Progressive Distillation and Retrospection (ECCV2018) [paper]
• PackNet: Adding Multiple Tasks to a Single Network by Iterative Pruning (CVPR2018) [paper] [code]
• Overcoming Catastrophic Forgetting with Hard Attention to the Task (ICML2018) [paper] [code]
• Lifelong Learning with Dynamically Expandable Networks (ICLR2018) [paper]
• FearNet: Brain-Inspired Model for Incremental Learning (ICLR2018) [paper]

2017

• Incremental Learning of Object Detectors Without Catastrophic Forgetting (ICCV2017) [paper]
• Overcoming catastrophic forgetting in neural networks (EWC) (PNAS2017) [paper] [code] [code]
• Continual Learning Through Synaptic Intelligence (ICML2017) [paper] [code]
• Gradient Episodic Memory for Continual Learning (NIPS2017) [paper] [code]
• iCaRL: Incremental Classifier and Representation Learning (CVPR2017) [paper] [code]
• Continual Learning with Deep Generative Replay (NIPS2017) [paper] [code]
• Overcoming Catastrophic Forgetting by Incremental Moment Matching (NIPS2017) [paper] [code]
• Expert Gate: Lifelong Learning with a Network of Experts (CVPR2017) [paper]
• Encoder Based Lifelong Learning (ICCV2017) [paper]

2016

• Learning without forgetting (ECCV2016) [paper] [code]

Awesome Long-Tailed Recognition / Imbalanced Learning

Find it interesting that there are more shared techniques than I thought for incremental learning (exemplars-based).

ContinualAI wiki

An Open Community of Researchers and Enthusiasts on Continual/Lifelong Learning for AI

Workshops

4th Lifelong Learning Workshop at ICML 2020

Workshop on Continual Learning at ICML 2020

Continual Learning in Computer Vision Workshop CVPR 2020

Continual learning workshop NeurIPS 2018

Challenges or Competitions

1st Lifelong Learning for Machine Translation Shared Task at WMT20 (EMNLP 2020)

Continual Learning in Computer Vision Challenge CVPR 2020

Lifelong Robotic Vision Challenge IROS 2019

原文：https://github.com/xialeiliu/Awesome-Incremental-Learning