Email: thomasjdelaney [AT] ircn.jp
My research interests lie in the use of deep learning to model perception in humans, especially visual perception. Specifically, I am interested in using these models to investigate potential underlying causes for ASD symptoms. I am currently using deep convolutional neural networks to asses the effect of imbalances in excitation and inhibition on visual perception. I also plan to use this Hallucination Machine to manifest the effects of these imbalances in a virtual reality environment.
During my PhD, my research was based around modelling the responses of large populations of neurons. I implemented the ‘Spectral Rejection’ method to detect functional communities in correlation based networks of neurons. I also invented a novel method for modelling the number of active neurons in a population that also captures the association between the neurons in that population. Finally, I created a biophysical model for the fluorescence trace produced by the soma of a neuron containing a fluorescent calcium indicator.