The computational challenges of neuroinformatics
- Speaker: Dan Goodman (Harvard Medical School)
- Host: Viv Woods
- 21st May 2014 at 14:15 in K.B. L.T 1.3
Neuroinformatics - the use of computational methods for the analysis and simulation of neural data - is rapidly becoming one of the key challenges for research in neuroscience. Advances in the design and manufacture of neural probes means that neuroscientists will soon be able to generate vastly more data than they are able to analyse with current techniques. This new large scale data will also spur interest in larger scale models and simulations, which will require computational resources and expertise beyond what most neuroscience laboratories are capable of. I will argue that (1) the technical problems are solvable by utilising diverse, modern computational hardware such as general purpose graphics processing units, and (2) that it is possible to make this technology accessible to non-experts. I will present my work towards solving these problems in two software packages: KlustaKwik, for electrophysiological data analysis, and Brian, a neural network simulator that uses automatic code generation for multiple computational devices. Time permitting, I will do an interactive demo of Brian with audience participation.