Photonic Delay Systems as Machine Learning Implementations

Abstract

Nonlinear photonic delay systems present interesting implementation platforms for machine learning models. They can be extremely fast, offer great degrees of parallelism and potentially consume far less power than digital processors. So far they have been successfully employed for signal processing using the Reservoir Computing paradigm. In this paper we show that their range of applicability can be greatly extended if we use gradient descent with backpropagation through time on a model of the system to optimize the input encoding of such systems. We perform physical experiments that demonstrate that the obtained input encodings work well in reality, and we show that optimized systems perform significantly better than the common Reservoir Computing approach. The results presented here demonstrate that common gradient descent techniques from machine learning may well be applicable on physical neuro-inspired analog computers.

Cited by in corpus (13)

• Quantum Machine Learning
• Recent Advances in Physical Reservoir Computing: A Review
• High performance photonic reservoir computer based on a coherently driven passive cavity
• Tutorial: Photonic Neural Networks in Delay Systems
• Basic protocols in quantum reinforcement learning with superconducting circuits
• Online Training of an Opto-Electronic Reservoir Computer Applied to Real-Time Channel Equalisation
• Training Passive Photonic Reservoirs with Integrated Optical Readout
• CONDENSE: A Reconfigurable Knowledge Acquisition Architecture for Future 5G IoT
• Stability Analysis of Reservoir Computers Dynamics via Lyapunov Functions
• Deep Photonic Reservoir Computer for Speech Recognition
• Reservoir Computers with Random and Optimized Time-Shifts
• Reservoir Computers Modal Decomposition and Optimization
• Towards Deep Physical Reservoir Computing Through Automatic Task Decomposition And Mapping