A Spin Glass Model for the Loss Surfaces of Generative Adversarial Networks
| Autor: | Nicholas P. Baskerville, Jonathan P. Keating, Francesco Mezzadri, Joseph Najnudel |
|---|---|
| Rok vydání: | 2021 |
| Předmět: |
FOS: Computer and information sciences
Computer Science - Machine Learning Generative adversarial networks math-ph cs.LG FOS: Physical sciences 01 natural sciences math.PR Article 010305 fluids & plasmas Machine Learning (cs.LG) Random matrix theory math.MP Spin glasses 0103 physical sciences FOS: Mathematics cond-mat.dis-nn 010306 general physics cond-mat.stat-mech Mathematical Physics Condensed Matter - Statistical Mechanics Statistical Mechanics (cond-mat.stat-mech) Probability (math.PR) Statistical and Nonlinear Physics Deep learning Mathematical Physics (math-ph) Disordered Systems and Neural Networks (cond-mat.dis-nn) Condensed Matter - Disordered Systems and Neural Networks Mathematics - Probability Neural networks |
| Zdroj: | Journal of Statistical Physics Baskerville, N P, Keating, J P, Mezzadri, F & Najnudel, J 2022, ' A spin-glass model for the loss surfaces of generative adversarial networks ', Journal of Statistical Physics, vol. 186, no. 2, 29 . https://doi.org/10.1007/s10955-022-02875-w University of Bristol-PURE |
| ISSN: | 0022-4715 |
| DOI: | 10.1007/s10955-022-02875-w |
| Popis: | 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. 26 pages, 9 figures |
| Databáze: | OpenAIRE |
| Externí odkaz: |
|
Full text from SpringerLink