Online Continual Learning in Image Classification: An Empirical Survey

Abstract

Online continual learning for image classification studies the problem of learning to classify images from an online stream of data and tasks, where tasks may include new classes (class incremental) or data nonstationarity (domain incremental). One of the key challenges of continual learning is to avoid catastrophic forgetting (CF), i.e., forgetting old tasks in the presence of more recent tasks. Over the past few years, many methods and tricks have been introduced to address this problem, but many have not been fairly and systematically compared under a variety of realistic and practical settings. To better understand the relative advantages of various approaches and the settings where they work best, this survey aims to (1) compare state-of-the-art methods such as MIR, iCARL, and GDumb and determine which works best at different experimental settings; (2) determine if the best class incremental methods are also competitive in domain incremental setting; (3) evaluate the performance of 7 simple but effective trick such as "review" trick and nearest class mean (NCM) classifier to assess their relative impact. Regarding (1), we observe iCaRL remains competitive when the memory buffer is small; GDumb outperforms many recently proposed methods in medium-size datasets and MIR performs the best in larger-scale datasets. For (2), we note that GDumb performs quite poorly while MIR -- already competitive for (1) -- is also strongly competitive in this very different but important setting. Overall, this allows us to conclude that MIR is overall a strong and versatile method across a wide variety of settings. For (3), we find that all 7 tricks are beneficial, and when augmented with the "review" trick and NCM classifier, MIR produces performance levels that bring online continual learning much closer to its ultimate goal of matching offline training.

arXiv commentAccepted for publication in the Elsevier's Neurocomputing journal. Codes available at https://github.com/RaptorMai/online-continual-learning

References in corpus (43)

• Overcoming catastrophic forgetting in neural networks
• Model-Agnostic Meta-Learning for Fast Adaptation of Deep Networks
• Riemannian Walk for Incremental Learning: Understanding Forgetting and Intransigence
• Continual Learning Through Synaptic Intelligence
• An Empirical Investigation of Catastrophic Forgetting in Gradient-Based Neural Networks
• PathNet: Evolution Channels Gradient Descent in Super Neural Networks
• Efficient Lifelong Learning with A-GEM
• Three scenarios for continual learning
• On Tiny Episodic Memories in Continual Learning
• Progress & Compress: A scalable framework for continual learning
• Overcoming Catastrophic Forgetting by Incremental Moment Matching
• Overcoming catastrophic forgetting with hard attention to the task
• Generative replay with feedback connections as a general strategy for continual learning
• Online Structured Laplace Approximations For Overcoming Catastrophic Forgetting
• Towards Robust Evaluations of Continual Learning
• LAMOL: LAnguage MOdeling for Lifelong Language Learning
• A Neural Dirichlet Process Mixture Model for Task-Free Continual Learning
• Dark Experience for General Continual Learning: a Strong, Simple Baseline
• Continual Unsupervised Representation Learning
• Coresets via Bilevel Optimization for Continual Learning and Streaming
• GraphSAIL: Graph Structure Aware Incremental Learning for Recommender Systems
• Dynamic Few-Shot Visual Learning without Forgetting
• Task Agnostic Continual Learning Using Online Variational Bayes
• Effects of Degradations on Deep Neural Network Architectures
• Deep Online Learning via Meta-Learning: Continual Adaptation for Model-Based RL
• Online Continual Learning on Sequences
• Rotate your Networks: Better Weight Consolidation and Less Catastrophic Forgetting
• La-MAML: Look-ahead Meta Learning for Continual Learning
• Multi-Task Incremental Learning for Object Detection
• End-to-End Incremental Learning
• Feature Transfer Learning for Deep Face Recognition with Under-Represented Data
• Lifelong and Interactive Learning of Factual Knowledge in Dialogues
• Continual Learning in Practice
• Batch-level Experience Replay with Review for Continual Learning
• Semantic Drift Compensation for Class-Incremental Learning
• Incremental Learning In Online Scenario
• Generative Feature Replay For Class-Incremental Learning
• Continual Learning with Knowledge Transfer for Sentiment Classification
• Continual Learning for Sentence Representations Using Conceptors
• Energy-Based Models for Continual Learning
• Continual Learning for Natural Language Generation in Task-oriented Dialog Systems
• ADER: Adaptively Distilled Exemplar Replay Towards Continual Learning for Session-based Recommendation
• Neuromodulated Neural Architectures with Local Error Signals for Memory-Constrained Online Continual Learning