Deep Photonic Reservoir Computing with On-chip Nonlinearity

TL;DR

This paper introduces a scalable, all-optical deep photonic reservoir computing system using silicon microring resonators for nonlinearity, achieving high accuracy and efficiency in spatiotemporal tasks with minimal training.

Contribution

The work presents a novel deep photonic reservoir computing framework that leverages on-chip silicon microring resonators for nonlinearity, enabling scalable, high-performance optical processing.

Findings

Achieves superior action recognition accuracy on NTU RGB D benchmark.

Demonstrates a high computational density of 334.25 TOPs/mm2.

Scales effectively with additional wavelength channels and reservoir layers.

Abstract

Reservoir computing, renowned for its low training cost, has emerged as a promising lightweight paradigm for efficient spatiotemporal processing,it remains challenging to realize deep photonic reservoir computing (DPRC) systems, due to the lack of scalable on-chip nonlinearity. Here, we introduce a versatile time delayed DPRC framework that natively supports deep and concurrent spatiotemporal processing entirely in the optical domain. At its core, the system leverages free carrier dynamics in silicon microring resonators to provide the fundamental nonlinearity and short term memory, and these nonlinear nodes are interconnected through true time delay lines that establish shared long-term memory. Benefiting from intrinsic physical nonlinearity and multi-timescale fading memory, this simple yet effective architecture demonstrates remarkable high dimensional representation capabilities. On the NTU RGB D benchmark, the parameter efficient DPRC system achieves superior action recognition accuracies compared to mainstream deep learning models, while requiring only a single shot regression training procedure. We further verify a prototype DPRC chip that excels across diverse dataset classification and time series prediction tasks. It enables a streamlined all optical pipeline between hierarchical layers, delivering a consistent computational density of 334.25 TOPs/mm2, independent of the reservoir depth and three orders of magnitude higher than conventional approaches. Moreover, its performance scales with near-zero hardware overhead by utilizing additional wavelength channels. This DPRC network is highly scalable on a silicon photonic platform, with flexible extension to hundreds of deep reservoir layers and parallel channels, paving the way toward intelligent optoelectronic systems for advanced real time processing and parallel decision making.