Saved in:
Bibliographic Details
Main Authors: Tsikouras, Nikos, Pantis, Yorgos, Mitliagkas, Ioannis, Tzamos, Christos
Format: Preprint
Published: 2025
Subjects:
Online Access:https://arxiv.org/abs/2510.19382
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866913142709157888
author Tsikouras, Nikos
Pantis, Yorgos
Mitliagkas, Ioannis
Tzamos, Christos
author_facet Tsikouras, Nikos
Pantis, Yorgos
Mitliagkas, Ioannis
Tzamos, Christos
contents Understanding the dynamics of feature learning in neural networks (NNs) remains a significant challenge. The work of (Mousavi-Hosseini et al., 2023) analyzes a multiple index teacher-student setting and shows that a two-layer student attains a low-rank structure in its first-layer weights when trained with stochastic gradient descent (SGD) and a strong regularizer. This structural property is known to reduce sample complexity of generalization. Indeed, in a second step, the same authors establish algorithm-specific learning guarantees under additional assumptions. In this paper, we focus exclusively on the structure discovery aspect and study it under weaker assumptions, more specifically: we allow (a) NNs of arbitrary size and depth, (b) with all parameters trainable, (c) under any smooth loss function, (d) tiny regularization, and (e) trained by any method that attains a second-order stationary point (SOSP), e.g.\ perturbed gradient descent (PGD). At the core of our approach is a key $\textit{derandomization}$ lemma, which states that optimizing the function $\mathbb{E}_{\mathbf{x}} \left[g_θ(\mathbf{W}\mathbf{x} + \mathbf{b})\right]$ converges to a point where $\mathbf{W} = \mathbf{0}$, under mild conditions. The fundamental nature of this lemma directly explains structure discovery and has immediate applications in other domains including an end-to-end approximation for MAXCUT, and computing Johnson-Lindenstrauss embeddings.
format Preprint
id arxiv_https___arxiv_org_abs_2510_19382
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle A Derandomization Framework for Structure Discovery: Applications in Neural Networks and Beyond
Tsikouras, Nikos
Pantis, Yorgos
Mitliagkas, Ioannis
Tzamos, Christos
Machine Learning
Understanding the dynamics of feature learning in neural networks (NNs) remains a significant challenge. The work of (Mousavi-Hosseini et al., 2023) analyzes a multiple index teacher-student setting and shows that a two-layer student attains a low-rank structure in its first-layer weights when trained with stochastic gradient descent (SGD) and a strong regularizer. This structural property is known to reduce sample complexity of generalization. Indeed, in a second step, the same authors establish algorithm-specific learning guarantees under additional assumptions. In this paper, we focus exclusively on the structure discovery aspect and study it under weaker assumptions, more specifically: we allow (a) NNs of arbitrary size and depth, (b) with all parameters trainable, (c) under any smooth loss function, (d) tiny regularization, and (e) trained by any method that attains a second-order stationary point (SOSP), e.g.\ perturbed gradient descent (PGD). At the core of our approach is a key $\textit{derandomization}$ lemma, which states that optimizing the function $\mathbb{E}_{\mathbf{x}} \left[g_θ(\mathbf{W}\mathbf{x} + \mathbf{b})\right]$ converges to a point where $\mathbf{W} = \mathbf{0}$, under mild conditions. The fundamental nature of this lemma directly explains structure discovery and has immediate applications in other domains including an end-to-end approximation for MAXCUT, and computing Johnson-Lindenstrauss embeddings.
title A Derandomization Framework for Structure Discovery: Applications in Neural Networks and Beyond
topic Machine Learning
url https://arxiv.org/abs/2510.19382