Computational metrics and parameters of an injection-locked large area semiconductor laser for neural network computing

Abstract

Artificial neural networks have become a staple computing technique in many fields. Yet, they present fundamental differences with classical computing hardware in the way they process information. Photonic implementations of neural network architectures potentially offer fundamental advantages over their electronic counterparts in terms of speed, processing parallelism, scalability and energy efficiency. Scalable and high performance photonic neural networks (PNNs) have been demonstrated, yet they remain scarce. In this work, we study the performance of such a scalable, fully parallel and autonomous PNN based on a large area vertical-cavity surface-emitting laser (LA-VCSEL). We show how the performance varies with different physical parameters, namely, injection wavelength, injection power, and bias current. Furthermore, we link these physical parameters to the general computational measures of consistency and dimensionality. We present a general method of gauging dimensionality in high dimensional nonlinear systems subject to noise, which could be applied to many systems in the context of neuromorphic computing. Our work will inform future implementations of spatially multiplexed VCSEL PNNs.

GHz Rate Neuromorphic Photonic Spiking Neural Network With a Single Vertical-Cavity Surface-Emitting Laser (VCSEL)

“…Here, the light output from different points on the surface was spatially-multiplexed creating the individual network nodes, which were interconnected via surface carrier interactions and optical diffraction. This LA-VCSEL-based system achieved low error performance during a 3-bit binary header classification task and a further report was generated studying the ideal conditions for reservoir operation [31]. Overall, these reports demonstrate that VCSELs offer an exciting platform for photonic RC systems using the inherent nonlinear dynamical behaviours produced in these devices, yielding high performance across diverse complex tasks (nonlinear channel equalisation, time series prediction, etc.)…”

“…Instead, laser-based RC systems use the inherent nonlinear dynamical responses of lasers (VCSELs in this case) subject to external optical injection and/or feedback to process information. Recently, VCSEL-based RC systems have been demonstrated on two different experimental architectures, namely time delay reservoirs (TDRs) [29], [30], [13], [14] and spatial-temporal reservoirs [15], [31]. Approaches based on the TDR architecture (shown in Fig.…”