An Accelerated LIF Neuronal Network Array for a Large Scale Mixed-Signal Neuromorphic Architecture

TL;DR

This paper introduces an accelerated, highly tunable LIF neuron array integrated into a neuromorphic chip, enabling large-scale, mixed-signal neural network processing with demonstrated cortical-like functions.

Contribution

It presents a novel, scalable LIF neuron array for neuromorphic hardware with extensive tunability and demonstrated cortical processing capabilities.

Findings

Neurons exhibit dynamics matching the LIF model.

Successful implementation of a winner-take-all network.

Calibration and measurement validate the design.

Abstract

We present an array of leaky integrate-and-fire (LIF) neuron circuits designed for the second-generation BrainScaleS mixed-signal 65-nm CMOS neuromorphic hardware. The neuronal array is embedded in the analog network core of a scaled-down prototype HICANN-DLS chip. Designed as continuous-time circuits, the neurons are highly tunable and reconfigurable elements with accelerated dynamics. Each neuron integrates input current from a multitude of incoming synapses and evokes a digital spike event output. The circuit offers a wide tuning range for synaptic and membrane time constants, as well as for refractory periods to cover a number of computational models. We elucidate our design methodology, underlying circuit design, calibration and measurement results from individual sub-circuits across multiple dies. The circuit dynamics match with the behavior of the LIF mathematical model. We further demonstrate a winner-take-all network on the prototype chip as a typical element of cortical processing.