Abstract:Artificial neural networks (ANNs) represent a fundamentally connectionnist and distributed approach to computing, and as such they differ from classical computers that utilize the von Neumann architecture. This has revived research interest in new unconventional hardware to enable more efficient implementations of ANNs rather than emulating them on traditional machines. In order to fully leverage the capabilities of this new generation of ANNs, optimization algorithms that take into account hardware limitations and imperfections are necessary. Photonics represents a particularly promising platform, offering scalability, high speed, energy efficiency, and the capability for parallel information processing. Yet, fully fledged implementations of autonomous optical neural networks (ONNs) with in-situ learning remain scarce. In this work, we propose a ternary weight architecture high-dimensional semiconductor laser-based ONN. We introduce a simple method for achieving ternary weights with Boolean hardware, significantly increasing the ONN's information processing capabilities. Furthermore, we design a novel in-situ optimization algorithm that is compatible with, both, Boolean and ternary weights, and provide a detailed hyperparameter study of said algorithm for two different tasks. Our novel algorithm results in benefits, both in terms of convergence speed and performance. Finally, we experimentally characterize the long-term inference stability of our ONN and find that it is extremely stable with a consistency above 99\% over a period of more than 10 hours, addressing one of the main concerns in the field. Our work is of particular relevance in the context of in-situ learning under restricted hardware resources, especially since minimizing the power consumption of auxiliary hardware is crucial to preserving efficiency gains achieved by non-von Neumann ANN implementations.

Countering Variations and Thermal Effects for Accurate Optical Neural Networks.

Robustness of Neuromorphic Computing with RRAM-based Crossbars and Optical Neural Networks.

Optical Analog Computing Devices Designed by Deep Neural Network

On-chip 4F-System Based on Concave Mirrors for Optical Neural Networks

Efficient On-Chip Learning for Optical Neural Networks Through Power-Aware Sparse Zeroth-Order Optimization

Design of optical neural networks with component imprecisions

Electrically Programmable Phase-Change Photonic Memory for Optical Neural Networks with Nanoseconds in Situ Training Capability

Optical neural networks: progress and challenges

Advances and Marine Applications of Optical Neural Network

Strategies for Training Optical Neural Networks

Annealing-inspired training of an optical neural network with ternary weights

Space-efficient optical computing with an integrated chip diffractive neural network

Optical Neural Ordinary Differential Equations

Digital-analog hybrid matrix multiplication processor for optical neural networks

A manufacture-friendly design framework for optical neural networks

Adversarial Attacks on an Optical Neural Network

Integrated multi-operand optical neurons for scalable and hardware-efficient deep learning

Parity-time symmetric optical neural networks

Control-free and efficient integrated photonic neural networks via hardware-aware training and pruning

Hardware-software co-design of slimmed optical neural networks.