The Benefits of Reusing Batches for Gradient Descent in Two-Layer Networks: Breaking the Curse of Information and Leap Exponents

Yatin Dandi; Emanuele Troiani; Luca Arnaboldi; Luca Pesce; Lenka Zdeborová; Florent Krzakala

2024 ICML ICML 2024

The Benefits of Reusing Batches for Gradient Descent in Two-Layer Networks: Breaking the Curse of Information and Leap Exponents

Abstract

No abstract available. No abstract was captured for this paper when its proceedings were scraped - most often an older edition published without one online. How the corpus is built.

Authors

Yatin Dandi, Emanuele Troiani, Luca Arnaboldi, Luca Pesce, Lenka Zdeborová, Florent Krzakala

Topics

Machine Learning > Optimization & Theory > Learning Theory Machine Learning > Optimization & Theory > Neural Network Optimization Deep Learning > Optimization & Theory > Optimization

Keywords

gradient descent information exponent dynamical mean field theory batch reuse leap exponent multi-pass

Download PDF