Slimmed optical neural networks with multiplexed neuron sets and a corresponding backpropagation training algorithm

TL;DR

The paper addresses the challenge of deploying WDM-based parallelism in ONNs when nonlinear activation induces inter-channel crosstalk. It introduces multiplexed neuron sets (MNS) implemented with SOAs and a backpropagation (BP) training algorithm that explicitly accounts for crosstalk via the full Jacobian of the nonlinear MNS block. The key contributions are a universal MNS framework for compressing multiple neurons into one device, a BP algorithm that mitigates crosstalk effects, and experimental-level evidence showing similar performance to traditional ONNs with a substantially reduced hardware footprint and energy consumption. This work provides a practical pathway toward scalable, energy-efficient WDM-ONNs and offers a general training strategy for nonlinear, crosstalk-affected photonic networks.

Abstract

Due to their intrinsic capabilities on parallel signal processing, optical neural networks (ONNs) have attracted extensive interests recently as a potential alternative to electronic artificial neural networks (ANNs) with reduced power consumption and low latency. Preliminary confirmation of the parallelism in optical computing has been widely done by applying the technology of wavelength division multiplexing (WDM) in the linear transformation part of neural networks. However, inter-channel crosstalk has obstructed WDM technologies to be deployed in nonlinear activation in ONNs. Here, we propose a universal WDM structure called multiplexed neuron sets (MNS) which apply WDM technologies to optical neurons and enable ONNs to be further compressed. A corresponding back-propagation (BP) training algorithm is proposed to alleviate or even cancel the influence of inter-channel crosstalk on MNS-based WDM-ONNs. For simplicity, semiconductor optical amplifiers (SOAs) are employed as an example of MNS to construct a WDM-ONN trained with the new algorithm. The result shows that the combination of MNS and the corresponding BP training algorithm significantly downsize the system and improve the energy efficiency to tens of times while giving similar performance to traditional ONNs.

Figures (9)

• Figure 1: (A) A scheme of a traditional FCNN; the layers are connected by the black lines, which corresponds to the weight matrix. The neurons separately realize the summation and nonlinear activation functions without influencing others. (B) An example of a nonlinear activation function and how it can be conceptually multiplexed in a single device.
• Figure 2: (A) A block diagram of a WDM-ONN with an MNS structure. Multiple neurons are encoded on various wavelengths and input into MNS. (B) The MNS structure in this work is realized by a multichannel SOA. (C) A schemed connection picture for a WDM-ONN with a hidden layer composed of MNS.
• Figure 3: For a 2-channel SOA. (A) Here, the output of Ch-2 versus the input of Ch-1 and input of Ch-2is visualized. The inset shows the overall gain versus the input of Ch-1 and the input of Ch-2. (B) The partial derivatives of the output to the inputs is visualized here: $\partial ({P_{out\_k}{[Ch\_2]}})/\partial ({P_{in\_k}}{[Ch\_1]})$. The partial derivatives of the output to the inputs is also visualized here:$\partial ({P_{out\_k}}{[Ch\_2]})/\partial ({P_{in\_k}}{[Ch\_2]})$.
• Figure 4: The term $\frac{{\partial {P_{out\_k}}}}{{\partial {P_{in\_i}}}}$ evaluates the crosstalk level brought by the ${i^{th}}$ channel. The x-axis and the y-axis are ${G_{ss}}$ and ${P_{sat}}$ respectively, which are the two parameters affect the crosstalk level. The red box indicates the origin of the inset on the right. It is obvious that the interchannel crosstalk level increases with ${G_{ss}}$.
• Figure 5: (A) The scheme of the proposed ONN for simulation. (B)-(D) The performance of the proposed ONN with 2-channel,4-channel and 6-channel multiplexing SOAs. The x-axis indicates the crosstalk level. The proposed ONN trained by the new BP algorithm demonstrates a steady performance as the crosstalk level and the number of multiplexed channels increases. The one trained by the traditional BP algorithm suffers performance degradation induced by interchannel crosstalk. (E)-(F) The performance improvement of the new BP algorithm over the traditional one rises as more channels of SOAs in the proposed ONN are multiplexed. The new BP algorithm shows significant relevance to larger ONN network with denser-multiplexed MNS structure.