Dr. Naren Ramakrishnan, Professor of Computer Science, VT, will give a talk entitled "Neural ODEs and their Applications in Scientific Machine Learning"

November 21, 2025

@10:10, 6-051 room, VTRC, Arlington (in-person), 440 Goodwin Hall, Blacksburg
For remote access, click here.

Abstract
Neural Ordinary Differential Equations (Neural ODEs) constitute an emerging approach for physical modeling using machine learning frameworks, diverging from traditional ML by attempting to directly encode and learn continuous dynamics in the model architecture. This talk will provide an accessible introduction to neural ODEs, illustrating how they generalize well-known machine learning models such as ResNets and also offer multiple, complementary, perspectives for understanding them through links to classical numerical solvers, reversible models, and implicit layers. This talk will also discuss applications in areas like: (i) epidemiology where neural ODEs can capture compartmental dynamics from time course data, and (ii) nuclear science, where they can be used for digital twin modeling.

Bio
Dr. Naren Ramakrishnan is the Thomas L. Phillips Professor of Engineering at Virginia Tech and leads AI and machine learning for the Institute for Advanced Computing. He directs the Sanghani Center for AI and Data Analytics and the Amazon-Virginia Tech Initiative in Efficient and Robust Machine Learning. His research interests span data science, forecasting, urban analytics, recommender systems, and computational epidemiology. He and his students have received more than ten Best Paper Awards at leading data mining, AI, and data science conferences and a "Test of Time" award at KDD'25. Ramakrishnan is a fellow of the ACM, AAAS, and IEEE.