DOLCIT Seminar

In-person attendance is limited.  Attendees joining in person must have a valid Caltech ID card. The Seminar will also be in Zoom (open to Caltech.edu accounts only). For the Zoom link, please contact [email protected].


Abstract:
Deep learning has disrupted nearly every major field of study from computer vision to genomics. The unparalleled success of these models has, in many cases, been fueled by an explosion of data. Millions of labeled images, thousands of annotated ICU admissions, and hundreds of hours of transcribed speech are common standards in the literature. Clinical neuroscience is a notable holdout to this trend. It is a field of unavoidably small datasets, massive patient variability, and complex (largely unknown) phenomena. My lab tackles these challenges across a spectrum of projects, from answering foundational neuroscientific questions to translational applications of neuroimaging data to exploratory directions for probing neural circuitry. One of our key strategies in this data-starved regime has been to blend the structure and interpretability of classical methods with the representational power of deep learning.

This talk will highlight three projects that span a range of "old meets new" methodologies. On the foundational front, I will discuss our work to predict complex behavioral deficits from brain connectivity data. This framework combines dictionary learning for structure-function integration with recurrent neural networks to parse the evolving brain states. Next, I will describe our translational work on epileptic seizure detection from multichannel EEG. We develop a probabilistic graphical model, where the latent variables capture the spatiotemporal spread of a seizure; they are complemented by a deep data likelihood model. Finally, I will touch on an exploratory initiative to inject emotional cues into human speech. Our approach combines diffeomorphic curve registration with generative adversarial networks.