Summary of Memory-efficient Continual Learning with Neural Collapse Contrastive, by Trung-anh Dang et al.
Memory-efficient Continual Learning with Neural Collapse Contrastive
by Trung-Anh Dang, Vincent Nguyen, Ngoc-Son Vu, Christel Vrain
First submitted to arxiv on: 3 Dec 2024
Categories
- Main: Computer Vision and Pattern Recognition (cs.CV)
- Secondary: Artificial Intelligence (cs.AI); Machine Learning (cs.LG)
GrooveSquid.com Paper Summaries
GrooveSquid.com’s goal is to make artificial intelligence research accessible by summarizing AI papers in simpler terms. Each summary below covers the same AI paper, written at different levels of difficulty. The medium difficulty and low difficulty versions are original summaries written by GrooveSquid.com, while the high difficulty version is the paper’s original abstract. Feel free to learn from the version that suits you best!
| Summary difficulty | Written by | Summary |
|---|---|---|
| High | Paper authors | High Difficulty Summary Read the original abstract here |
| Medium | GrooveSquid.com (original content) | Medium Difficulty Summary The paper proposes Focal Neural Collapse Contrastive (FNC^2), a novel representation learning loss that balances both soft and hard relationships between samples. This addresses the issue of catastrophic forgetting in continual learning by leveraging the recently identified Neural Collapse phenomenon. The approach introduces Hardness-Softness Distillation (HSD) to preserve knowledge gained across tasks. Compared to state-of-the-art methods, FNC^2 minimizes memory reliance while rivaling rehearsal-based approaches without memory use, making it a promising solution for data privacy concerns. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary This paper tries to solve a big problem in artificial intelligence called “catastrophic forgetting”. This happens when machines learn from lots of different tasks and then forget old skills as they get new ones. The researchers propose a new way to balance two types of relationships between data samples, which helps the machine remember old skills better. Their approach works well even without using any special memory storage, making it useful for keeping personal data private. |
Keywords
» Artificial intelligence » Continual learning » Distillation » Representation learning
