Spiking Cerebellar Network Model: Effects of Long-Term Synaptic Plasticity and its Application in Brain-Inspired Control

Brain-inspired computing is a novel computing paradigm to realize the advanced artificial intelligence which is inspired by the working mechanism of human brain. As an important part of human brain, the cerebellum plays an important role in motor control and motor learning, which can help us explore more intelligent robot control methods. The motor control function of cerebellum depends on the mechanism of long-term synaptic plasticity. This paper focuses on the dynamic effects of long-term synaptic plasticity on cerebellar networks. The synaptic plasticity in cerebellum exists in three groups of cell junctions, which are from granule cells to purkinje (PKJ) cells, from mossy fiber to deep cerebellar nuclei, and from PKJ to deep cerebellar nuclei. Based on this physiological observation, we investigate the effects of these three mechanisms of synaptic plasticity on the firing rate of PKJ cells. Furthermore, we study the effects of three mechanisms of synaptic plasticity on brain-inspired control performance by adjusting the ratio of long-term potentiation (LTP) to long-term depression (LTD). The results show that PKJ neurons will have higher firing rates when the ratio of LTP to LTD is higher. When the synaptic plasticity mechanism of LTP plays a dominant role in control process, it will produce better control performance in brain-inspired control process. This research can be applied to intelligent robots, unmanned systems, edge computing of Internet of Things and brain-computer interface.