Reinforcement Learning for FPGA Spiking Neural Network coupled by LEDs <Abstract> In previous work, many neural network models such as Izhikevich model has been proposed until now. They describe neuron characteristic accurately. However it is one of barriers to apply many problem that the neural networks proposed in previous works have large calculation cost. To solve this problem, calculation on hardware is efficient way. This paper presents spiking neural network which is constructed by field programmable gate array (FPGA) and connected by Infrared Red (IR) communications with LEDs. In addition, implementation of reinforcement learning on this network and evaluation of the result are presented |