Presentation Information
[10p-N106-4]1M1C Reservoir Circuit Based on Pt/Nb:SrTiO3 Memristor with Regulatable Time Scale
〇(M2)Ryosuke Murai1, Yumeng Zheng1, Kentaro Kinoshita1 (1.Tokyo Univ. of Sci.)
Keywords:
physical reservoir computing,ReRAM,memristor
Physical reservoir computing is an effective machine learning model for learning time series information in terms of fast processing and low power consumption. However, the time scale for high learning performance is limited to around the time constant of the transient response of physical dynamics. In this study, we proposed 1M1C PRC system consisting of a Pt/Nb:SrTiO3 memristor with regulatable resistive state and a capacitor, and evaluated its learning performance. The results show that the time scale can be regulated by a simple method in the proposed circuit.