Summary of Asymptotics Of Feature Learning in Two-layer Networks After One Gradient-step, by Hugo Cui et al.
Asymptotics of feature learning in two-layer networks after one gradient-step
by Hugo Cui, Luca Pesce, Yatin Dandi, Florent Krzakala, Yue M. Lu, Lenka Zdeborová, Bruno Loureiro
First submitted to arxiv on: 7 Feb 2024
Categories
- Main: Machine Learning (stat.ML)
- Secondary: Disordered Systems and Neural Networks (cond-mat.dis-nn); Machine Learning (cs.LG)
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 This paper investigates how two-layer neural networks learn features from data and improve their performance after a single training step. Building on previous research, the authors develop a spiked Random Features (sRF) model to describe the trained network. They then provide an exact characterization of the generalization error in the high-dimensional limit, where the number of samples, width, and input dimension grow together. This allows them to understand how the network adapts to the data to efficiently learn non-linear functions. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary This paper looks at how neural networks learn from small amounts of training data. It’s like trying to understand how a person learns new skills after just one try. The authors create a special kind of model that helps explain how this happens. They find that the network is able to adapt and learn quickly because it’s able to change its behavior based on the data it sees. This is important for making neural networks work better in real-world situations. |
Keywords
* Artificial intelligence * Generalization
