Summary of Flattened One-bit Stochastic Gradient Descent: Compressed Distributed Optimization with Controlled Variance, by Alexander Stollenwerk and Laurent Jacques
Flattened one-bit stochastic gradient descent: compressed distributed optimization with controlled variance
by Alexander Stollenwerk, Laurent Jacques
First submitted to arxiv on: 17 May 2024
Categories
- Main: Machine Learning (cs.LG)
- Secondary: Numerical Analysis (math.NA); 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 The novel algorithm proposed for distributed stochastic gradient descent (SGD) with compressed gradient communication in the parameter-server framework leverages two simple ideas: one-bit quantization and a randomized fast Walsh-Hadamard transform. The resulting FO-SGD algorithm provides biased gradient approximations, preventing issues like exploding variance and sparse gradients. The technique is shown to offer SGD-like convergence guarantees under mild conditions and can be used for both worker-server communication directions. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary The paper proposes a new way to train machine learning models on large datasets by reducing the amount of data that needs to be sent between computers. It uses two clever techniques: one-bit compression and a special type of transformation called the Walsh-Hadamard transform. This helps solve problems like exploding variance, where small errors can add up quickly. The method is shown to work well and can be used in many different scenarios. |
Keywords
» Artificial intelligence » Machine learning » Quantization » Stochastic gradient descent
