Speaker: Alex Williams
Title: Statistical methods to characterize variability and individuality in neural recordings
Abstract: As in many scientific fields, datasets in neuroscience have reached a scale that defies manual inspection and visualization without the aid of advanced computational and mathematical tools. My recently established lab at NYU and the Flatiron Institute broadly aims to develop these tools. We are particularly interested in using statistical models to characterize flexibility and variability in neural circuits—e.g., how do the dynamics of large neural ensembles change over the course of learning a new skill, during periods of high attention or task engagement, or over the course of development and aging. In this talk, I will first provide an overview of these questions and the wide range of approaches we have used to address them, including tensor-structured factor models, time warping models, and Bayesian nonparametric methods to perform inference in cluster point processes. I will conclude by highlighting a recent area of interest: the use of statistical shape analysis to compare neural circuit dynamics across different animals (and also to compare hidden layer representations across deep neural networks).
FI employees are welcome to attend in person. Please email email@example.com for the Zoom link if you wish to attend remotely.
Visitors (w/out an FI badge) please email firstname.lastname@example.org to be registered to the building or to obtain Zoom information. All visitors must be vaccinated for entry into our buildings.