[8p-N304-1~14] FS.1 Focused Session "AI Electronics"

5:45 PM - 6:00 PM JST(8:45 AM - 9:00 AM UTC)

[8p-N304-14] Structural Parameter Analysis of Higher-Order Information Processing Capacity in Simple Cycle Reservoirs

〇Shimon Matsuno¹, Kota Ando¹, Tetsuya Asai¹ (1.Hokkaido Univ.)

Keywords:

Reservoir Computing

In edge AI, it is required to process sensor time-series data near the terminal under low power consumption. In this study, we show that for a reservoir computing model with a simple ring structure, the Simple Cycle Reservoir, the balance between memory of past inputs and nonlinear computational capability can be tuned by only a small number of parameters that characterize the connection pattern, providing guidelines for designing energy-efficient hardware.