Skip to main navigation Skip to search Skip to main content

Dense Associative Memory in a nonlinear-optical Hopfield Neural Network

  • Stevens Institute of Technology

Research output: Chapter in Book/Report/Conference proceedingConference contributionpeer-review

Abstract

We present an experimental realization of a nonlinear-optical Hopfield neural network (NOHNN) designed for high-capacity dense associative memory (DAM). By exploiting the intrinsic nonlinearity of second-harmonic generation (SHG), we physically implement effective 4-body neuron interactions that significantly exceed the storage limits of traditional 2-body systems. The NOHNN demonstrates a minimum tenfold improvement in the storage of uncorrelated patterns and up to a fifty-fold enhancement for correlated patterns. Benchmarking using MNIST handwritten digits reveals a 5.5 times increase in storage capacity together with improved noise tolerance. These results highlight nonlinear photonics as a promising hardware platform for scalable associative memory and optimization.

Original languageEnglish
Title of host publicationComputational Optical Imaging and Artificial Intelligence in Biomedical Sciences III
EditorsLiang Gao, Guoan Zheng, Seung Ah Lee
ISBN (Electronic)9781510696433
DOIs
StatePublished - 5 Mar 2026
Event2026 3rd Computational Optical Imaging and Artificial Intelligence in Biomedical Sciences - San Francisco, United States
Duration: 17 Jan 202620 Jan 2026

Publication series

NameProgress in Biomedical Optics and Imaging - Proceedings of SPIE
Volume13865
ISSN (Print)1605-7422
ISSN (Electronic)2410-9045

Conference

Conference2026 3rd Computational Optical Imaging and Artificial Intelligence in Biomedical Sciences
Country/TerritoryUnited States
CitySan Francisco
Period17/01/2620/01/26

Keywords

  • Dense Associative Memory
  • Hopfield Network
  • Nonlinear Optics
  • Optical Neural Networks
  • Photonic Computing
  • Second-Harmonic Generation

Fingerprint

Dive into the research topics of 'Dense Associative Memory in a nonlinear-optical Hopfield Neural Network'. Together they form a unique fingerprint.

Cite this