Summary of Rehearsal-free Federated Domain-incremental Learning, by Rui Sun et al.
Rehearsal-free Federated Domain-incremental Learning
by Rui Sun, Haoran Duan, Jiahua Dong, Varun Ojha, Tejal Shah, Rajiv Ranjan
First submitted to arxiv on: 22 May 2024
Categories
- Main: Machine Learning (cs.LG)
- Secondary: Computer Vision and Pattern Recognition (cs.CV)
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 A novel federated domain incremental learning framework, RefFiL, is proposed to address the challenges of catastrophic forgetting in federated learning. The method leverages a global prompt-sharing paradigm and learns domain-invariant knowledge, incorporating prompts from different domains represented by FL participants. A key feature is the generation of local fine-grained prompts by a domain adaptive prompt generator, which maintains distinctive boundaries on a global scale. A contrastive learning loss is also introduced to differentiate between locally generated prompts and those from other domains, enhancing RefFiL’s precision. The approach significantly alleviates catastrophic forgetting without requiring extra memory space, making it suitable for privacy-sensitive devices. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary RefFiL is a new way for computers to learn together while remembering what they learned before. This is important because when we try to teach computers new things, they often forget the old things. In this case, RefFiL uses a special kind of sharing to help the computers remember. It creates unique “prompts” that are specific to each group of devices learning together. This helps the computers learn from each other without forgetting what they knew before. |
Keywords
» Artificial intelligence » Federated learning » Precision » Prompt
