Summary of Distributed Learning Based on 1-bit Gradient Coding in the Presence Of Stragglers, by Chengxi Li and Mikael Skoglund
Distributed Learning based on 1-Bit Gradient Coding in the Presence of Stragglers
by Chengxi Li, Mikael Skoglund
First submitted to arxiv on: 19 Mar 2024
Categories
- Main: Machine Learning (cs.LG)
- Secondary: Distributed, Parallel, and Cluster Computing (cs.DC)
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 (DL) method called 1-bit gradient coding (1-bit GCDL) is proposed to reduce the communication burden in DL systems. The existing gradient coding-based approaches require workers to transmit real-valued vectors, resulting in high communication overhead. To address this issue, 1-bit GCDL encodes locally computed gradients into 1-bit data and transmits them to reduce the communication cost. Convergence guarantees are theoretically provided for both convex and non-convex loss functions. Experimental results show that 1-bit GC-DL outperforms baseline methods under the same communication overhead. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary This paper explores a way to make machines learn together without using too much internet bandwidth. When computers work together, they need to share information with each other. This takes up a lot of space on the internet. The researchers created a new way for computers to send this information that uses less data, called 1-bit gradient coding. They tested this method and found it was better than existing methods at learning while using less bandwidth. |
