Summary of Stochastic Gradient Descent For Streaming Linear and Rectified Linear Systems with Adversarial Corruptions, by Halyun Jeong et al.
Stochastic gradient descent for streaming linear and rectified linear systems with adversarial corruptions
by Halyun Jeong, Deanna Needell, Elizaveta Rebrova
First submitted to arxiv on: 2 Mar 2024
Categories
- Main: Machine Learning (cs.LG)
- Secondary: Numerical Analysis (math.NA); Machine Learning (stat.ML)
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 SGD-exp approach addresses robustness concerns in linear and ReLU regressions under Massart noise, a semi-random corruption model. This paper provides novel guarantees for nearly linear convergence to the true parameter with up to 50% corruption rate or symmetric oblivious corruptions. The improved convergence rate over previous methods is attributed to an exponentially decaying step size, which has been shown to be efficient in practice. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary This research shows a new way to train models that are resistant to noise and errors. By using a special kind of gradient descent, the model can learn even when some data points are corrupted or missing. The method is tested on linear and ReLU regressions and shows good results. This means that it could be useful for real-world applications where data is noisy or incomplete. |
Keywords
* Artificial intelligence * Gradient descent * Relu
