Summary of Convex Formulations For Training Two-layer Relu Neural Networks, by Karthik Prakhya et al.
Convex Formulations for Training Two-Layer ReLU Neural Networks
by Karthik Prakhya, Tolga Birdal, Alp Yurtsever
First submitted to arxiv on: 29 Oct 2024
Categories
- Main: Machine Learning (cs.LG)
- Secondary: 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 Machine learning educators who are not specialized in the subfield of infinite-width two-layer ReLU networks will appreciate this paper’s novel approach to training neural networks. The authors reformulate the problem as a convex completely positive program, which is NP-hard due to the complete positivity constraint. To overcome this challenge, they introduce a semidefinite relaxation that can be solved in polynomial time. The paper experimentally evaluates the tightness of this relaxation, demonstrating its competitive performance in test accuracy across various classification tasks. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary This paper helps us make better machine learning models by solving big optimization problems. Right now, we don’t fully understand how some models work inside. That’s because they’re too complex and hard to solve. The authors found a new way to solve these problems that makes them easier to understand and train. They tested this new method and showed it works well for classifying things into different categories. |
Keywords
» Artificial intelligence » Classification » Machine learning » Optimization » Relu
