[16p-M_123-1~8] FS.1 Focused Session "AI Electronics"

4:00 PM - 4:15 PM JST(7:00 AM - 7:15 AM UTC)

[16p-M_123-8] Computational Properties Induced by Reconfigurable Topology in Reservoir Computing

〇(M2)Shimon Matsuno¹, Kota Ando¹, Tetsuya Asai¹ (1.Hokkaido Univ.)

Keywords:

Reservoir Computing,Physical Reservoir Computing

This study focuses on a Simple Cycle Reservoir (SCR) with a simple structure suitable for hardware implementation and investigates how topology reconfiguration affects its information processing properties. To address the limited nonlinearity of SCRs, a configuration that selectively switches a small number of shortcut connections is introduced. This approach enables modulation of nonlinearity without significantly altering the overall structure and provides design guidelines for dynamic performance control using reconfigurable topologies.