Design of optical neural networks with component imprecisions

12/13/2019
by   Michael Y. -S. Fang, et al.
0

For the benefit of designing scalable, fault resistant optical neural networks (ONNs), we investigate the effects architectural designs have on the ONNs' robustness to imprecise components. We train two ONNs – one with a more tunable design (GridNet) and one with better fault tolerance (FFTNet) – to classify handwritten digits. When simulated without any imperfections, GridNet yields a better accuracy ( 98 amount of error in their photonic components, the more fault tolerant FFTNet overtakes GridNet. We further provide thorough quantitative and qualitative analyses of ONNs' sensitivity to varying levels and types of imprecisions. Our results offer guidelines for the principled design of fault-tolerant ONNs as well as a foundation for further research.

READ FULL TEXT

page 4

page 7

page 8

page 9

page 18

page 19

page 20

page 21

research
02/17/2022

Winograd Convolution: A Perspective from Fault Tolerance

Winograd convolution is originally proposed to reduce the computing over...
research
09/11/2019

Parallel fault-tolerant programming of an arbitrary feedforward photonic network

Reconfigurable photonic mesh networks of tunable beamsplitter nodes can ...
research
03/25/2021

Actuator Fault-Tolerant Vehicle Motion Control: A Survey

The advent of automated vehicles operating at SAE levels 4 and 5 poses h...
research
04/06/2022

A Design Methodology for Fault-Tolerant Computing using Astrocyte Neural Networks

We propose a design methodology to facilitate fault tolerance of deep le...
research
12/30/2021

A Survey of fault mitigation techniques for multi-core architectures

Fault tolerance in multi-core architecture has attracted attention of re...
research
10/02/2019

Overview of Fault Tolerant Techniques in Underwater Sensor Networks

Sensor networks provide services to a broad range of applications rangin...
research
07/06/2019

Adversarial Fault Tolerant Training for Deep Neural Networks

Deep Learning Accelerators are prone to faults which manifest in the for...

Please sign up or login with your details

Forgot password? Click here to reset