Nonlinear classification of neural manifolds with contextual information

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Mignacco, Francesca, Chou, Chi-Ning, Chung, SueYeon
Format: Preprint
Published: 2024
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866916665325780992
author Mignacco, Francesca
Chou, Chi-Ning
Chung, SueYeon
author_facet Mignacco, Francesca
Chou, Chi-Ning
Chung, SueYeon
contents Understanding how neural systems efficiently process information through distributed representations is a fundamental challenge at the interface of neuroscience and machine learning. Recent approaches analyze the statistical and geometrical attributes of neural representations as population-level mechanistic descriptors of task implementation. In particular, manifold capacity has emerged as a promising framework linking population geometry to the separability of neural manifolds. However, this metric has been limited to linear readouts. To address this limitation, we introduce a theoretical framework that leverages latent directions in input space, which can be related to contextual information. We derive an exact formula for the context-dependent manifold capacity that depends on manifold geometry and context correlations, and validate it on synthetic and real data. Our framework's increased expressivity captures representation reformatting in deep networks at early stages of the layer hierarchy, previously inaccessible to analysis. As context-dependent nonlinearity is ubiquitous in neural systems, our data-driven and theoretically grounded approach promises to elucidate context-dependent computation across scales, datasets, and models.
format Preprint
id arxiv_https___arxiv_org_abs_2405_06851
institution arXiv
publishDate 2024
record_format arxiv
spellingShingle Nonlinear classification of neural manifolds with contextual information
Mignacco, Francesca
Chou, Chi-Ning
Chung, SueYeon
Neurons and Cognition
Disordered Systems and Neural Networks
Statistical Mechanics
Neural and Evolutionary Computing
Machine Learning
Understanding how neural systems efficiently process information through distributed representations is a fundamental challenge at the interface of neuroscience and machine learning. Recent approaches analyze the statistical and geometrical attributes of neural representations as population-level mechanistic descriptors of task implementation. In particular, manifold capacity has emerged as a promising framework linking population geometry to the separability of neural manifolds. However, this metric has been limited to linear readouts. To address this limitation, we introduce a theoretical framework that leverages latent directions in input space, which can be related to contextual information. We derive an exact formula for the context-dependent manifold capacity that depends on manifold geometry and context correlations, and validate it on synthetic and real data. Our framework's increased expressivity captures representation reformatting in deep networks at early stages of the layer hierarchy, previously inaccessible to analysis. As context-dependent nonlinearity is ubiquitous in neural systems, our data-driven and theoretically grounded approach promises to elucidate context-dependent computation across scales, datasets, and models.
title Nonlinear classification of neural manifolds with contextual information
topic Neurons and Cognition
Disordered Systems and Neural Networks
Statistical Mechanics
Neural and Evolutionary Computing
Machine Learning
url https://arxiv.org/abs/2405.06851