[16p-A25-1~16] 3.14 Silicon photonics and integrated photonics

4:30 PM - 4:45 PM JST(7:30 AM - 7:45 AM UTC)Presentation by Applicant for JSAP Young Scientists Presentation Award

[16p-A25-11] Leaky integrated-and-fire model for photonic spiking neuron with magnetic materials

〇Gaku Takagi¹, Yuya Shoji¹ (1.TokyoTech.)

Keywords:

magneto-optical effect,magnetization reversal,spiking neural network

Optical accelerators are attracting attention as computing devices capable of processing advanced AI models with high speed and low power consumption. Research has been reported on fully-photonic neurons using phase-change materials (PCM), but a problem arises due to the limitation on the number of times PCM can be rewritten by optical pulses. To overcome this limitation, we propose an fully-photonic spiking neuron using magneto-optical materials. We experimentally verified that the thermal absorption response of magnetic materials is similar to the leaky integrate-and-fire model of spiking neurons.