A spin-glass model for the loss surfaces of generative adversarial networks

DOI10.1007/S10955-022-02875-WMaRDI QIDQ6357780arXivFDO

Authors Nicholas P. Baskerville, J. P. Keating, F. Mezzadri, Joseph Najnudel

Publication date 7 January 2021

Published in arXiv

Full work available at URL https://doi.org/10.1007/S10955-022-02875-W

Mathematics Subject Classification ID

Special matrices (15Bxx) Time-dependent statistical mechanics (dynamic and nonequilibrium) (82Cxx) Applications of statistical mechanics to specific types of physical systems (82Dxx)

arXiv classification

cond-mat.dis-nncond-mat.stat-mechcs.LGmath-phmath.MPmath.PR

Abstract: We present a novel mathematical model that seeks to capture the key design feature of generative adversarial networks (GANs). Our model consists of two interacting spin glasses, and we conduct an extensive theoretical analysis of the complexity of the model's critical points using techniques from Random Matrix Theory. The result is insights into the loss surfaces of large GANs that build upon prior insights for simpler networks, but also reveal new structure unique to this setting.

This page was built for publication: A spin-glass model for the loss surfaces of generative adversarial networks

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q6357780)