Surya Ganguli, Stanford University

High Dimensional Single Trial Neural Data Analysis: Theory and Practice

Remarkable advances in experimental neuroscience now enable us to simultaneously observe the activity of many neurons for long periods of time, thereby providing an opportunity to understand how the moment by moment collective dynamics of the brain instantiates cognition and behavior, and how longer term changes in these dynamics mediate learning. However, efficiently extracting such a conceptual understanding from large, high dimensional neural datasets requires concomitant advances in theoretically driven experimental design and data analysis. We will discuss how the development of modern theories and algorithms in high dimensional statistics can be exploited to (1) find optimal ways to discover patterns in high dimensional data, (2) tradeoff very different experimental resources, like numbers of recorded neurons and trials to discover latent cognitive states, and (3) extract unbiased and interpretable low dimensional descriptions of how rapid single-trial circuit dynamics change slowly over many trials to mediate learning.

Additional Information

Symposium talk; registration is necessary

Organized by

BCCN Berlin