Bokun Zhao ^1,2* Xuening Dong ^1,2 Kaveh Rahbardar Mojaver ^1,2 Brett H. Meyer ^1,2 Odile Liboiron-Ladouceur ^1,2

• ¹ Faculty of Engineering, McGill University, Montreal, Canada
• ² McGill University, Montreal, Quebec, Canada

Optical neural networks implemented with Mach-Zehnder Interferometer (MZI) arrays are a promising solution to enable fast and energy-efficient machine learning inference, yet finding a practical application has proven challenging due to sensitivity to thermal noise and loss. In this work, we propose the binary optical trigger as a promising field of application for the current state of optical computing. Implementable as small-scale application-specific circuitry on edge devices, binary triggers leverage the distinct advantages of integrated optical processors while avoiding its shortcomings given the current state of optical computing. Motivated by the limited task complexity, constrained area and power budgets of binary triggers, we perform systematic application-specific hardware pruning by physically removing specific MZIs, resulting in a customized MZI-mesh topology whose structure provides adequate performance for a targeted task complexity. We demonstrate in simulation that the pruning methodology achieves at least 50% less MZI usage compared to Clements and Reck meshes with the same input size, translating to at least between 4.6% and 24.2% savings in power consumption and a 40% reduction in physical circuitry footprint compared to other proposed unitary MZI topologies, sacrificing only 1% to 2% drop in inference accuracy.