Photonic Neuromorphic Accelerator for Convolutional Neural Networks based on an Integrated Reconfigurable Mesh

Fuente: arXiv
Salvato in:
Dettagli Bibliografici
Autori principali: Tsirigotis, Aris, Sarantoglou, Gerge, Deligiannidis, Stavros, Sanchez, Erica, Gutierrez, Ana, Bogris, Adonis, Capmany, Jose, Mesaritakis, Charis
Natura: Preprint
Pubblicazione: 2024
Soggetti:
Accesso online:
Tags: Aggiungi Tag
Nessun Tag, puoi essere il primo ad aggiungerne!!
_version_ 1866917662751195136
author Tsirigotis, Aris
Sarantoglou, Gerge
Deligiannidis, Stavros
Sanchez, Erica
Gutierrez, Ana
Bogris, Adonis
Capmany, Jose
Mesaritakis, Charis
author_facet Tsirigotis, Aris
Sarantoglou, Gerge
Deligiannidis, Stavros
Sanchez, Erica
Gutierrez, Ana
Bogris, Adonis
Capmany, Jose
Mesaritakis, Charis
contents In this work, we present and experimentally validate a passive photonic-integrated neuromorphic accelerator that uses a hardware-friendly optical spectrum slicing technique through a reconfigurable silicon photonic mesh. The proposed scheme acts as an analogue convolutional engine, enabling information preprocessing in the optical domain, dimensionality reduction and extraction of spatio-temporal features. Numerical results demonstrate that utilizing only 7 passive photonic nodes, critical modules of a digital convolutional neural network can be replaced. As a result, a 98.6% accuracy on the MNIST dataset was achieved, with a power consumption reduction of at least 26% compared to digital CNNs. Experimental results confirm these findings, achieving 97.7% accuracy with only 3 passive nodes.
format Preprint
id arxiv_https___arxiv_org_abs_2405_06434
institution arXiv
publishDate 2024
record_format arxiv
spellingShingle Photonic Neuromorphic Accelerator for Convolutional Neural Networks based on an Integrated Reconfigurable Mesh
Tsirigotis, Aris
Sarantoglou, Gerge
Deligiannidis, Stavros
Sanchez, Erica
Gutierrez, Ana
Bogris, Adonis
Capmany, Jose
Mesaritakis, Charis
Optics
Image and Video Processing
In this work, we present and experimentally validate a passive photonic-integrated neuromorphic accelerator that uses a hardware-friendly optical spectrum slicing technique through a reconfigurable silicon photonic mesh. The proposed scheme acts as an analogue convolutional engine, enabling information preprocessing in the optical domain, dimensionality reduction and extraction of spatio-temporal features. Numerical results demonstrate that utilizing only 7 passive photonic nodes, critical modules of a digital convolutional neural network can be replaced. As a result, a 98.6% accuracy on the MNIST dataset was achieved, with a power consumption reduction of at least 26% compared to digital CNNs. Experimental results confirm these findings, achieving 97.7% accuracy with only 3 passive nodes.
title Photonic Neuromorphic Accelerator for Convolutional Neural Networks based on an Integrated Reconfigurable Mesh
topic Optics
Image and Video Processing
url https://arxiv.org/abs/2405.06434