Summary of Personalized Federated Continual Learning Via Multi-granularity Prompt, by Hao Yu et al.
Personalized Federated Continual Learning via Multi-granularity Prompt
by Hao Yu, Xin Yang, Xin Gao, Yan Kang, Hao Wang, Junbo Zhang, Tianrui Li
First submitted to arxiv on: 27 Jun 2024
Categories
- Main: Machine Learning (cs.LG)
- Secondary: Artificial Intelligence (cs.AI)
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 proposed Personalized Federated Continual Learning (PFCL) framework addresses the challenges of sharing and personalizing knowledge in multi-granularity representation. It combines knowledge fusion for server aggregation with model improvement for each client, aiming to overcome Spatial-Temporal Catastrophic Forgetting (STCF). The novel concept of multi-granularity prompts is introduced, consisting of coarse-grained global prompts and fine-grained local prompts. The former focuses on transferring shared global knowledge without spatial forgetting, while the latter emphasizes specific learning of personalized local knowledge to overcome temporal forgetting. A selective prompt fusion mechanism is designed for aggregating knowledge from different clients. Experimental results demonstrate the effectiveness of the proposed method in addressing STCF and improving personalized performance. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary PFCL is a new way to share and personalize knowledge. It’s like trying to learn something new, but everyone has their own unique perspective. The problem is that when people try to learn together, they often forget what they learned before or don’t remember the big picture. PFCL tries to solve this by combining two things: sharing global knowledge with others and personalizing learning for each individual. It uses a special type of “prompt” that helps people remember and build on what they’ve learned before, while also learning new things. |
Keywords
» Artificial intelligence » Continual learning » Prompt
