A Comprehensive Study of Class Incremental Learning Algorithms for Visual Tasks

Abstract

The ability of artificial agents to increment their capabilities when confronted with new data is an open challenge in artificial intelligence. The main challenge faced in such cases is catastrophic forgetting, i.e., the tendency of neural networks to underfit past data when new ones are ingested. A first group of approaches tackles forgetting by increasing deep model capacity to accommodate new knowledge. A second type of approaches fix the deep model size and introduce a mechanism whose objective is to ensure a good compromise between stability and plasticity of the model. While the first type of algorithms were compared thoroughly, this is not the case for methods which exploit a fixed size model. Here, we focus on the latter, place them in a common conceptual and experimental framework and propose the following contributions: (1) define six desirable properties of incremental learning algorithms and analyze them according to these properties, (2) introduce a unified formalization of the class-incremental learning problem, (3) propose a common evaluation framework which is more thorough than existing ones in terms of number of datasets, size of datasets, size of bounded memory and number of incremental states, (4) investigate the usefulness of herding for past exemplars selection, (5) provide experimental evidence that it is possible to obtain competitive performance without the use of knowledge distillation to tackle catastrophic forgetting and (6) facilitate reproducibility by integrating all tested methods in a common open-source repository. The main experimental finding is that none of the existing algorithms achieves the best results in all evaluated settings. Important differences arise notably if a bounded memory of past classes is allowed or not.

Accepted for publication in the Elsevier's Neural Networks journal

References in corpus (24)

• Scikit-learn: Machine Learning in Python
• Distilling the Knowledge in a Neural Network
• A systematic study of the class imbalance problem in convolutional neural networks
• Progressive Neural Networks
• CNN Features off-the-shelf: an Astounding Baseline for Recognition
• Learning multiple visual domains with residual adapters
• Born Again Neural Networks
• FearNet: Brain-Inspired Model for Incremental Learning
• Continual Learning with Deep Generative Replay
• Towards Understanding Knowledge Distillation
• Continual Learning in Generative Adversarial Nets
• Incremental Classifier Learning with Generative Adversarial Networks
• Relational Knowledge Distillation
• M2KD: Multi-model and Multi-level Knowledge Distillation for Incremental Learning
• A Strategy for an Uncompromising Incremental Learner
• Piggyback: Adapting a Single Network to Multiple Tasks by Learning to Mask Weights
• End-to-End Incremental Learning
• REMIND Your Neural Network to Prevent Catastrophic Forgetting
• Few-Shot Class-Incremental Learning
• Semantic Drift Compensation for Class-Incremental Learning
• Lifelong Machine Learning with Deep Streaming Linear Discriminant Analysis
• Incremental Learning with Maximum Entropy Regularization: Rethinking Forgetting and Intransigence
• DeeSIL: Deep-Shallow Incremental Learning
• GAN Memory with No Forgetting

Cited by in corpus (24)

• Class-Incremental Learning: A Survey
• Continual Learning and Catastrophic Forgetting
• A Survey on Continual Semantic Segmentation: Theory, Challenge, Method and Application
• Lifelong Continual Learning for Anomaly Detection: New Challenges, Perspectives, and Insights
• Inherit with Distillation and Evolve with Contrast: Exploring Class Incremental Semantic Segmentation Without Exemplar Memory
• Self-Supervised Training Enhances Online Continual Learning
• Replay in Deep Learning: Current Approaches and Missing Biological Elements
• Balanced softmax cross-entropy for incremental learning with and without memory
• Learning at a Glance: Towards Interpretable Data-limited Continual Semantic Segmentation via Semantic-Invariance Modelling
• A First Look at Class Incremental Learning in Deep Learning Mobile Traffic Classification
• FeTrIL: Feature Translation for Exemplar-Free Class-Incremental Learning
• Self-Supervised Features Improve Open-World Learning
• Hyperdimensional Representation Learning for Node Classification and Link Prediction
• Contrastive Supervised Distillation for Continual Representation Learning
• On the importance of cross-task features for class-incremental learning
• CL2R: Compatible Lifelong Learning Representations
• Detecting Unknown DGAs without Context Information
• Future-Proofing Class-Incremental Learning
• Memory-Free Generative Replay For Class-Incremental Learning
• Class-wise Classifier Design Capable of Continual Learning using Adaptive Resonance Theory-based Topological Clustering
• REAL: Representation Enhanced Analytic Learning for Exemplar-free Class-incremental Learning
• Multiple Queries with Multiple Keys: A Precise Prompt Matching Paradigm for Prompt-based Continual Learning
• Selective Replay Enhances Learning in Online Continual Analogical Reasoning
• Disentangling Transfer and Interference in Multi-Domain Learning