[16p-A33-1~16] FS.1 Focused Session "AI Electronics"

2:00 PM - 2:15 PM JST(5:00 AM - 5:15 AM UTC)

[16p-A33-2] Realtime handwriting anomaly detection using an FPGA-based spiking reservoir

〇Hisashi Inoue¹, Hiroto Tamura², Ai Kitoh¹, Xiangyu Chen³, Zolboo Byambadorj³, Takeaki Yajima⁴, Yasushi Hotta⁵, Tetsuya Iizuka³, Gouhei Tanaka^2,6, Isao Inoue¹ (1.AIST, 2.Univ. Tokyo (IRCN), 3.Univ. Tokyo (Dept. Engineering), 4.Kyushu Univ., 5.Univ. Hyogo, 6.Nagoya Inst. Tech.)

Keywords:

reservoir computing,spiking neural network,anomaly detection

Reservoir comuting is a promising neural network-based computing technology that excels at processing time-series data at high energy efficiency. To evaluate energy efficiency, we have developed a spiking neural network-based reservoir using FPGA-based digital hardware. Using the handwriting anomaly detection of handwritten symbols as a model task of reservoir computing, we discuss the detection accuracy and energy consumption of this task.