Summary of Fedccl: Federated Dual-clustered Feature Contrast Under Domain Heterogeneity, by Yu Qiao et al.
FedCCL: Federated Dual-Clustered Feature Contrast Under Domain Heterogeneity
by Yu Qiao, Huy Q. Le, Mengchun Zhang, Apurba Adhikary, Chaoning Zhang, Choong Seon Hong
First submitted to arxiv on: 14 Apr 2024
Categories
- Main: Computer Vision and Pattern Recognition (cs.CV)
- 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 Federated learning (FL) enables decentralized neural network training through collaboration between edge clients and a central server, addressing privacy concerns. One challenge is the non-Ideal distributed data, characterized by both intra-domain and inter-domain heterogeneity. To address this, researchers have employed averaged signals as regularization, focusing on a single aspect of these challenges. This paper clarifies the two non-IID challenges and proposes a novel framework for cluster representation-based FL that tackles these issues from local and global perspectives. The proposed dual-clustered feature contrast-based FL framework consists of two stages: intra-class information capture through local clustering and cross-client knowledge sharing via global clustering and averaging. Experimental results on multiple datasets demonstrate the effectiveness of this approach in achieving comparable or superior performance gains under heterogeneity. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary This paper is about a way to train artificial neural networks without collecting all the data in one place, while keeping it private. This is important because some data might be sensitive or belong to different people. The problem is that the data from different places (like homes or cities) isn’t the same and can’t be easily combined. To solve this, researchers are trying new methods. This paper proposes a way to group similar data together at both local and global levels, allowing for more accurate training while keeping sensitive information private. The results show that this method performs well on various datasets. |
Keywords
» Artificial intelligence » Clustering » Federated learning » Neural network » Regularization
