Title of the talk: "  Learning Dynamics: From Neural ODE Generalization to Data-Driven Disease Modeling  "

Date & Time: Nov 11,2025 at 05:00 PM(ONLINE)

Abstract: This talk bridges theoretical and applied perspectives in learning dynamical systems. In the first part, I will discuss our recent work on generalization capacities of Neural Ordinary Differential Equations (Neural ODEs), where we derive generalization error bounds for time-dependent and time-independent cases with general nonlinear dynamics. The analysis reveals how Lipschitz continuity, overparameterization, and domain bounds influence model generalization. In the second part, I will present a data-driven framework for modeling malaria transmission dynamics, integrating temperature- and altitude-dependent transmission functions within a compartmental SIR–SI model. We further employ physics-informed neural networks (PINNs) for physics-constrained parameter estimation, and use Dynamic Mode Decomposition (DMD) to derive a real-time, interpretable transmission risk index. Together, these studies demonstrate how theory-driven insights and data-driven modeling can jointly enhance our understanding of complex dynamical systems. 

About the Speaker: Dr. Manoj is working as an Assistant Professor in the School of Engineering and Science at IIT Madras, Zanzibar since October 2023. His current research focuses on the modeling and analysis of infectious diseases using a combination of physics-informed machine learning, deep learning, and graph neural networks (GNNs). 

He has completed his Ph.D. from the School of Mathematical and Statistical Sciences, Indian Institute of Technology Mandi (IIT Mandi) and master's from the Indian Institute of Technology Ropar (IIT Ropar) in 2018.