The Japan Society of Applied Physics

[PS-5-07] Nanophotonic Neural Network and Generic Algorism Computing Using n-InP/Electro-Optic Polymer Hybrid Modulator

Tomoki Sakuma1, Shiyoshi Yokoyama2, Junichi Fujikata3 (1. Tokushima Univ. (Japan), 2. Inst. of Materials Chemistry, Kyushu Univ. (Japan), 3. Inst. of post-LED photonics, Tokushima Univ. (Japan))

https://doi.org/10.7567/SSDM.2023.PS-5-07

Artificial neural networks have been attracting attention in recent years, and high-speed operation with low power consumption are required. Therefore, we used InP/electro-optic polymer hybrid modulation for optical neural networks to solve that problem. In addition, we used genetic algorithm to learn that is tolerant for optical loss in the hybrid optical modulator.