Neuromorphic Engineering

neuromorphic engineeringEfficient, parallel, low-power computation is a hallmark of brain computation and the goal of neuromorphic engineering. The focus of this project is to design, implement and test the most accurate, electronic, very large scale integrated (VLSI) circuit model of the cochlea and its associated auditory signal processing.

In creating this electronic model, MARCS Institute for Brain, Behaviour and Development will develop new schemes for parallel, low-power, auditory signal processing that would be impossible to study in any other way. The cochlea model will accurately simulate the fluid dynamic properties of the biological cochlea and will include active gain control and active quality factor control of the cochlea partition. It will also implement the processing performed by the sensory transducers and the spiking neurons of the auditory nerve.

In addition to modelling the cochlea, we will also create spiking neuron integrated circuits capable of simulating the response properties of many of the types found in biology. These will enable us to implement the computations underlying low level auditory perception in real time. Their implementations will be directly informed by the computational models developed in computational modelling of the human brain.

A new standard exists, the 'Address Event Representation' protocol, to which not only our implementations will adhere, but also many existing and future integrated circuits designed by other neuromorphic engineers. This interface is furthermore programmable, enabling it to perform computation on the spike trains as they pass through the interface from one chip to another. This tool enables advanced spike-based signal processing systems.

In addition to the electronic approach, we will also aim to build a real-time software model of the auditory pathway. The software model will have the advantage of being flexible and does not suffer from mismatch or noise. The disadvantage is that for a computer simulation to run in real time on a computer, simplifications must necessarily be made, such as ignoring many of the non-linear behaviours of the cochlea and neural processing, or reduction of the frequency range and resolution over which the system operates.