Fundamental computational limits of weak learnability in high-dimensional multi-index models

TL;DR

This paper investigates the theoretical limits of efficiently learning low-dimensional structures in high-dimensional multi-index models, revealing phase transitions and hierarchical learning phenomena relevant to neural network training.

Contribution

It characterizes the sample complexity thresholds and phase transitions for weakly recovering subspaces in multi-index models, extending understanding of neural network learnability.

Findings

Identifies conditions for trivial subspace learnability with minimal samples.

Establishes a phase transition at a critical sample complexity for learning non-trivial subspaces.

Discovers hierarchical learning where directions are learned sequentially based on difficulty.

Abstract

Multi-index models - functions which only depend on the covariates through a non-linear transformation of their projection on a subspace - are a useful benchmark for investigating feature learning with neural nets. This paper examines the theoretical boundaries of efficient learnability in this hypothesis class, focusing on the minimum sample complexity required for weakly recovering their low-dimensional structure with first-order iterative algorithms, in the high-dimensional regime where the number of samples $n\!=\!\alpha d$ is proportional to the covariate dimension $d$. Our findings unfold in three parts: (i) we identify under which conditions a trivial subspace can be learned with a single step of a first-order algorithm for any $\alpha\!>\!0$; (ii) if the trivial subspace is empty, we provide necessary and sufficient conditions for the existence of an easy subspace where directions that can be learned only above a certain sample complexity $\alpha\!>\!\alpha_c$, where $\alpha_{c}$ marks a computational phase transition. In a limited but interesting set of really hard directions -- akin to the parity problem -- $\alpha_c$ is found to diverge. Finally, (iii) we show that interactions between different directions can result in an intricate hierarchical learning phenomenon, where directions can be learned sequentially when coupled to easier ones. We discuss in detail the grand staircase picture associated to these functions (and contrast it with the original staircase one). Our theory builds on the optimality of approximate message-passing among first-order iterative methods, delineating the fundamental learnability limit across a broad spectrum of algorithms, including neural networks trained with gradient descent, which we discuss in this context.