Akai Lab.

Reservoir computation is a neural network computation that is expected to become a new type of artificial intelligence.
Since 2018, research on physically creating this computation has attracted a great deal of attention worldwide. Studies in this field have encompassed not only solid-state elements, but also spin, light, and various other materials. Our laboratory has elucidated that the chemical reactions and electrical transient responses generated at the electrode surfaces in aqueous solution can solve the nonlinear problems required for reservoir calculations. The signals were simultaneously measured at multiple terminals as diverse electrochemical currents in distilled water and aqueous acid molecular solutions with high redox reactions.
As we continue our research utilizing chemical reactions at surfaces for informational processing, we intend to focus our attention on the electrochemical reactions brought about by ions and the nonlinear responses of ionic molecules and dynamics in their complex systems.

• Akai-Kasaya M. et al., " Performance of reservoir computing in a random network of single-walled carbon nanotubes complexed with polyoxometalate",Neuromorphic Computing and Engineering, vol. 2, (2022)
• Kan S. et al., "Physical implementation of reservoir computing through electrochemical reaction",Advanced Science, vol. 9, pp. 2104076(1)-2104076(8) (2021).
• 赤井 恵, 浅井 哲也, 中嶋 浩平, "学習する有機材料：リザバー計算に向けたニューラルネットワーク形成," 応用物理, vol. 90, no. 8, pp. 504-508 (2021).
• Kan S. et al., "Simple reservoir computing capitalizing on the nonlinear response of materials: Theory and physical implementations," Physical Review Applied, vol. 15, pp. 024030-1-024030-12 (2021).