[18p-438-1~11] 3.10 Optical quantum physics and technologies

△ [18p-438-4] Optical reservoir computing with combination of reinforcement learning

〇Kazutaka Kanno¹, Makoto Naruse², Atsushi Uchida¹ (1.Saitama Univ., 2.NICT)

Keywords:reservoir computing, reinforcement learning, laser

In prediction problems, the prediction acuracy of a prediction model may be degraeded when a prediction target changes. For solving above problem, we propose optical reservoir computing with combination of reinforcement learning. In the proposed method, some prediction models are trained for different prediction target. Their models are selected using reinforcement learning. It is shown in our numerical simulation that a proper prediction model can be selected with reinforcement learning in a time-series prediction task where time-series generated from different targets temporally switch.