Spike Neural Network of Motor Cortex Model for Arm Reaching Controlling
Motor Cortex modeling is significant to understand movement planning and execution, and inter-connected recurrent neural network has successful described the dynamics of neurons. However, most of existing methods take continues signal based neural networks to simulate the MC activities, which do not reflect the biological spike neural signal. To address this limitation, we proposed a spike neural network based modeling approach to model the Motor Cortex activity in arm reaching task. Specifically, we built a model of motor cortex based on integrate-and-fire spiking neurons with conductance-based synapses. Second, we carefully design the interconnections of neurons with two different time scale and assigned them into two layers. Besides, prior knowledge based parameter initialization is also proposed to ensure the spikes communication and the training performance. Experiments show the effectiveness of proposed the method that high similar with real spike at single cell and population level.