Learning in Silicon Beyond STDP: A Neuromorphic Implementation of Multi-Factor Synaptic Plasticity With Calcium-Based Dynamics

Abstract

Autonomous systems must be able to adapt to a constantly-changing environment. This adaptability requires significant computational resources devoted to learning, and current artificial systems are lacking in these resources when compared to humans and animals. We aim to produce VLSI spiking neural networks which feature learning structures similar to those in biology, with the goal of achieving the performance and efficiency of natural systems. The neuroscience literature suggests that calcium ions play a key role in explaining long-term synaptic plasticity's dependence on multiple factors, such as spike timing and stimulus frequency. Here we present a novel VLSI implementation of a calcium-based synaptic plasticity model, comparisons between the model and circuit simulations, and measurements of the fabricated circuit.