# Synaptic Delays for Temporal Feature Detection in Dynamic Neuromorphic Processors

## TL;DR

This paper explores disynaptic delay elements in neuromorphic processors inspired by cricket auditory circuits, demonstrating their potential for temporal feature detection and how they can be tuned via synaptic parameters.

## Contribution

It introduces a novel implementation of disynaptic delay elements in neuromorphic hardware and analyzes their behavior and robustness in mimicking biological auditory processing.

## Key findings

- Disynaptic delay elements can be configured to produce controlled timing of delayed excitation.
- Device mismatch affects the distribution of delays but can be managed through configuration.
- The delay mechanism is robust to input noise and temperature variations.

## Abstract

Spiking neural networks implemented in dynamic neuromorphic processors are well suited for spatiotemporal feature detection and learning, for example in ultra low-power embedded intelligence and deep edge applications. Such pattern recognition networks naturally involve a combination of dynamic delay mechanisms and coincidence detection. Inspired by an auditory feature detection circuit in crickets, featuring a delayed excitation by postinhibitory rebound, we investigate disynaptic delay elements formed by inhibitory-excitatory pairs of dynamic synapses. We configure such disynaptic delay elements in the DYNAP-SE neuromorphic processor and characterize the distribution of delayed excitations resulting from device mismatch. Furthermore, we present a network that mimics the auditory feature detection circuit of crickets and demonstrate how varying synapse weights, input noise and processor temperature affects the circuit. Interestingly, we find that the disynaptic delay elements can be configured such that the timing and magnitude of the delayed postsynaptic excitation depend mainly on the efficacy of the inhibitory and excitatory synapses, respectively. Delay elements of this kind can be implemented in other reconfigurable dynamic neuromorphic processors and opens up for synapse level temporal feature tuning with large fan-in and flexible delays of order 10-100 ms.