Doctoral Thesis: Generative models for structured neural time series
Doctoral Candidate: Andrew Song
- Date: Friday, November 5
- Time: 12pm
- Location: 34-401A
Additional Location Details:
Abstract: In neural signal processing, the fundamental assumption is that the signals from seemingly complex brain are not purely random, but rather those that have latent structures that can
be recovered with principled approaches. In this talk, I define this framework from the Bayesian/optimization perspective and emphasize translating and integrating the clinical and scientific domain knowledge, obtained from constant interaction/collaboration with the experimental neuroscientists and clinicians.
I will focus on uncovering latent structures in the neural time series data, by using domain priors/constraints, such as Gaussian process, shift-invariance, sparsity, and smoothness, among many others.
I demonstrate that the Bayesian approach with careful integration of these constraints produces results/structures in the data that are not only interpretable but also better performing for the metrics of interest.
Bio: Andrew H. Song is a final year Ph.D. student at MIT EECS, currently being co-advised by Emery N. Brown (MIT) and Demba Ba (Harvard). His current research interest is in the intersection of statistical signal processing and computational neuroscience. Before starting Ph.D. at MIT, he completed undergraduate degree in EECS at MIT.
- Dr. Emery Brown