Neural modelling on SpiNNaker

This project aims will deliver more complex neuron models, such as the two-compartment cortical pyramidal cell, on the SpiNNaker neuromorphic computing platform. It will achieve this by exploiting the shared resources provided by the system and a more efficient mapping of neural algorithms onto it. What the student wants to achieve is to demonstrate the capabilities of the SpiNNaker platform to handle more sophisticated models than the current point-neuron models, while maintaining the machine's energy-efficient characteristics, and to explore how far is possible to proceed along this path. This research fits within novel computing, neuriomorphic and many-core computing agendas, with potential applications in both brain science and event-based AI systems. The work requires knowledge of multicore computing and shared memory management. The first phase of the research is an analysis of the neuron models listed in the literature. The chosen models will then be implemented on the SpiNNaker platform and complexity, perfomance and power analysis undertaken.