This repository serves as a curated collection of academic papers, code, and resources focused on Incremental Learning (also known as Lifelong Learning or Continual Learning). It aims to provide a comprehensive overview of the field, covering theoretical advancements, algorithmic approaches, and practical applications for researchers and practitioners in machine learning.

How It Works

The repository organizes a vast number of research papers, primarily from top-tier AI conferences and journals, categorized by year and topic. It highlights key surveys, seminal works, and recent advancements in areas like catastrophic forgetting, knowledge distillation, replay mechanisms, and applications across computer vision, NLP, and reinforcement learning. The collection emphasizes both foundational concepts and cutting-edge techniques.

Quick Start & Requirements

This is a curated list of papers and does not have a direct installation or execution command. Users can browse the extensive list of papers, many of which include links to their respective code repositories (often PyTorch or TensorFlow).

Highlighted Details

• Extensive coverage of papers from 2016 to the present, with a significant focus on recent advancements (2023-2025).
• Includes links to associated code for many papers, facilitating reproducibility and practical experimentation.
• Covers a wide range of sub-fields within continual learning, including class-incremental, task-incremental, domain-incremental, and few-shot incremental learning.
• Features links to relevant workshops, challenges, and community resources like the ContinualAI wiki.

Maintenance & Community

The repository is maintained by xialeiliu and appears to be actively updated with new research. It encourages community contributions for missing papers.

Licensing & Compatibility

The repository itself is a collection of links to external research papers and code. The licensing and compatibility of individual papers and their associated code will vary based on their original publication and repository licenses.

Limitations & Caveats

As a curated list, the repository does not provide a unified framework or codebase for continual learning. Users must navigate to individual paper repositories for code and specific implementation details. The sheer volume of papers can be overwhelming without a clear starting point.