A Synaptically Local Spike-Rate-Dependent Plasticity toward Neuromorphic Computing

A synaptically local spike-rate-dependent learning rule is proposed in this work. Using mathematical analysis and a numerical simulation, we demonstrate that the synaptic weights weaken/strengthen with the proposed method when the presynaptic firing rate is low/high. Such changes correspond to the rate-dependent synaptic plasticity observed in actual neural circuits. We also present that the proposed method enables learning in a classical conditioning manner. As synaptically local spike-rate-dependent plasticity, our proposed method is expected to contribute to training spiking neural networks on rate-encoded information without complicating neuromorphic circuits.