Resources
Curated surveys, frameworks, benchmarks, and learning materials for continual learning research.
Surveys & Overviews
Continual Lifelong Learning with Neural Networks: A Review
2019Comprehensive survey covering the main CL paradigms, evaluation protocols, and open challenges.
A Comprehensive Study of Class Incremental Learning
2020In-depth study of class-incremental learning methods with a unified experimental framework.
Online Continual Learning in Image Classification
2021Survey focused on online CL settings, including single-pass and data-stream scenarios.
Frameworks & Libraries
Avalanche
End-to-end library for continual learning research built on PyTorch, featuring benchmarks, strategies, and evaluation tools.
Sequoia
A research framework for continual, transfer, and multi-task learning with a tree of settings.
ContinualAI
The largest research community and organization dedicated to continual learning in AI.
Benchmarks
Split CIFAR-100
CIFAR-100 split into sequential tasks — the most widely used benchmark for class-incremental learning.
Split MNIST / Permuted MNIST
Classic continual learning benchmarks for task-incremental and domain-incremental settings.
Split TinyImageNet
A more challenging benchmark derived from ImageNet with 200 classes and higher resolution images.
Tutorials & Courses
ContinualAI Wiki
Community-maintained wiki with in-depth explanations of continual learning concepts and methods.
CVPR Continual Learning Workshops
Annual workshops at CVPR featuring the latest advances in continual learning for computer vision.
Continual Learning Course (ContinualAI)
Free online course covering fundamentals of continual learning, from theory to practical implementations.