Summary of Communication Compression For Distributed Learning Without Control Variates, by Tomas Ortega et al.
Communication Compression for Distributed Learning without Control Variates
by Tomas Ortega, Chun-Yin Huang, Xiaoxiao Li, Hamid Jafarkhani
First submitted to arxiv on: 5 Dec 2024
Categories
- Main: Machine Learning (cs.LG)
- Secondary: Signal Processing (eess.SP); Optimization and Control (math.OC)
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 distributed learning framework, Compressed Aggregate Feedback (CAFe), is proposed to reduce communication costs in Federated Learning (FL) while preserving privacy. The framework allows for highly compressible client updates by leveraging past aggregated updates, eliminating the need for control variates and stateful clients. CAFe outperforms Distributed Compressed Gradient Descent (DCGD) with biased compression in non-smooth regimes with bounded gradient dissimilarity, according to theoretical proofs. Experimental results confirm CAFe’s superiority over direct compression methods and demonstrate its ability to compress client updates. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary In this paper, a new way of sharing information between devices is introduced. This method, called Compressed Aggregate Feedback (CAFe), helps reduce the amount of data sent between devices while keeping personal information private. It does this by using past information instead of sending everything from scratch. CAFe works better than other methods that try to compress data in certain situations. The authors tested CAFe and found it performs well. |
Keywords
» Artificial intelligence » Federated learning » Gradient descent
