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

Download:

Current browse context:

physics

Change to browse by:

References & Citations

Bookmark

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

Condensed Matter > Disordered Systems and Neural Networks

Title: Supervised perceptron learning vs unsupervised Hebbian unlearning: Approaching optimal memory retrieval in Hopfield-like networks

Abstract: The Hebbian unlearning algorithm, i.e. an unsupervised local procedure used to improve the retrieval properties in Hopfield-like neural networks, is numerically compared to a supervised algorithm to train a linear symmetric perceptron. We analyze the stability of the stored memories: basins of attraction obtained by the Hebbian unlearning technique are found to be comparable in size to those obtained in the symmetric perceptron, while the two algorithms are found to converge in the same region of Gardner's space of interactions, having followed similar learning paths. A geometric interpretation of Hebbian unlearning is proposed to explain its optimal performances. Because the Hopfield model is also a prototypical model of disordered magnetic system, it might be possible to translate our results to other models of interest for memory storage in materials.
Comments: 11 pages, 13 figures, 43 references
Subjects: Disordered Systems and Neural Networks (cond-mat.dis-nn); Statistical Mechanics (cond-mat.stat-mech); Biological Physics (physics.bio-ph); Neurons and Cognition (q-bio.NC)
Journal reference: J. Chem. Phys. 156, 104107 (2022)
DOI: 10.1063/5.0084219
Cite as: arXiv:2201.00032 [cond-mat.dis-nn]
  (or arXiv:2201.00032v2 [cond-mat.dis-nn] for this version)

Submission history

From: Enrico Ventura [view email]
[v1] Fri, 31 Dec 2021 19:16:08 GMT (938kb,D)
[v2] Mon, 7 Mar 2022 12:43:55 GMT (1021kb,D)

Link back to: arXiv, form interface, contact.