We gratefully acknowledge support from
the Simons Foundation and member institutions.
Full-text links:

Download:

Current browse context:

q-bio.NC

Change to browse by:

References & Citations

Bookmark

(what is this?)
CiteULike logo BibSonomy logo Mendeley logo del.icio.us logo Digg logo Reddit logo

Quantitative Biology > Neurons and Cognition

Title: Nonlinear classification of neural manifolds with contextual information

Abstract: 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. Here, we propose a theoretical framework that overcomes this limitation by leveraging contextual input information. We derive an exact formula for the context-dependent capacity that depends on manifold geometry and context correlations, and validate it on synthetic and real data. Our framework's increased expressivity captures representation untanglement 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.
Comments: 5 pages, 5 figures
Subjects: Neurons and Cognition (q-bio.NC); Disordered Systems and Neural Networks (cond-mat.dis-nn); Statistical Mechanics (cond-mat.stat-mech); Neural and Evolutionary Computing (cs.NE); Machine Learning (stat.ML)
Cite as: arXiv:2405.06851 [q-bio.NC]
  (or arXiv:2405.06851v1 [q-bio.NC] for this version)

Submission history

From: Francesca Mignacco [view email]
[v1] Fri, 10 May 2024 23:37:31 GMT (1799kb,D)

Link back to: arXiv, form interface, contact.