Seminar: Information Flow and Representation in Neural Circuits
Praveen Venkatesh, Ph.D. Candidate at Carnegie Mellon University presented an online seminar titled Information Flow and Representation in Neural Circuits.
You can view a recording of the seminar talk via our YouTube Channel
A central problem in neuroscience is to understand the representation and flow of information in healthy and diseased states of the brain. Such an understanding can help inform interventions to modulate information flows and treat brain diseases and disorders. The first part of my talk will provide a new framework for thinking about information flow in the brain, and propose a definition for information flow about a "message", e.g., the stimulus. Drawing from existing literature and simple examples, we will see that an important aspect of defining information flow in a meaningful way is to recognize and account for "synergistic" information representation. With this, I will show that it is possible to verifiably track information flow about a given message. To tackle the issue of synergistic representation, I leverage recent developments in information theory on "Partial Information Decomposition", which provides new tools for defining and estimating unique, redundant and synergistic information. In showing how these new measures can be utilized, I also explain how synergy can arise in unexpected ways in the brain, through a case study on grid cells. Finally, I will touch upon some of my work on improving neurotechnologies: investigating the fundamental limits of the spatial resolution of high-density EEG, and validating its efficacy through novel experiments.
Praveen Venkatesh received a B.Tech. (Hons.) in Electrical Engineering from the Indian Institute of Technology, Madras in 2014. He is currently pursuing a Ph.D. in Electrical and Computer Engineering at Carnegie Mellon University. Visit his professional page.
His work lies at the intersection of information theory, statistics, neuroscience and neuroengineering. He has collaborated with theorists and practitioners to identify fundamental theoretical problems, and addressed these using formal mathematical and computational models. His broad goal is to contribute to the development, study and application of new information-theoretic and statistical methods in neuroscience and neuroengineering. Over the course of his Ph.D., he has developed new frameworks and techniques for information flow in neuroscience, and studied the fundamental limits of high-density Electroencephalography (EEG). He has endeavoured to bring his theory to practice by collaborating closely with experimentalists and clinicians. Recently, he has also been working on using information-theoretic tools to bridge the fields of neuroscience and explainability in AI.
He is a recipient of the Henry L. Hillman Presidential Fellowship, the Dowd Fellowship from the College of Engineering, and the CMLH Fellowship in Digital Health from the Center for Machine Learning and Health at Carnegie Mellon University.