معرفی
Dibyakanti Kumar is a researcher in the Department of Computer Science, specializing in machine learning, computational mathematics, and fluid dynamics. His work focuses on neural networks, partial differential equations (PDEs), and their applications in solving complex physical systems. He has contributed to advancing theoretical guarantees for deep learning models and their application to fluid dynamics problems. His recent research includes developing data-driven methods for aerodynamic simulations and analyzing the stability of neural networks near critical phenomena like finite-time blow-up in PDEs.
Key research interests include:
- Neural network training and generalization
- Numerical methods for PDEs
- Physics-informed machine learning
- Data augmentation techniques
Recent academic activities include an oral presentation on the risks of solving fluid dynamics problems with neural networks, delivered in December 2024. His collaborations span interdisciplinary topics, integrating mathematical theory with engineering applications.
