Neuromorphic Circuits for Short-term Plasticity with Recovery Control

We present real-time neuromorphic VLSI circuits that implement the synaptic dynamics of Short Term Plasticity (STP). STP supports useful signal processing computational primitives such as change detection and gain control. Compact circuits implementing these mechanisms play a key role in providing neuromorphic VLSI systems with autonomous adaptation capabilities. We propose two different, flexible, short-term adaptation CMOS circuits for controlling the efficacy of synapses in response to incoming spikes. These circuits can be configured to either implement short-term depression or facilitation, with independent control over the adaptation and recovery rates. Our results demonstrate the dynamic properties of the proposed circuits and their behaviour in the frequency domain.