Events Detail

Speaker:
Timothy Dunn, Duke University

Join Zoom Meeting
zoom.us/j/81260587908

Meeting ID: 812 6058 7908

Timothy Dunn is a Forge Scholar and neuroscience researcher specializing in machine learning, particularly deep convolutional neural networks, as well as machine vision, and computational biology. His work has focused on how the brain controls behavior. His current postdoc research targets the new computational and theoretical challenges confronting systems neuroscientists. Dr. Dunn is an AI Watson XPrize Finalist (with team DataKind), the Grand Prize to be announced at TED 2020.

Geometric deep learning for 3D movement profiling across species and environments

Animal behavior is primarily conveyed by movement, and thus a basic understanding of the biology of behavior, and its control by the brain, hinges on the ability to measure how animals move. Yet despite the fundamental importance of movement, there has been a notable lack of tools for precise 3D movement quantification across scales, from motor primitives to coordinated behaviors of the individual and social groups. As such, current descriptions of animal behavior have been coarse and ad hoc, limiting investigation of the nervous system. My recent work bridges this gap, establishing experimental technologies and deep learning algorithms that provide high-resolution 3D readouts of animal behavior in mice, rats, marmosets, and chickadees. Further, I have used 3D movement analysis to resolve previously inaccessible brain states. My goal for this seminar is to foster discussion of future applications to exciting questions in ethology, neurobiology, and ecology.