Mehul Goyal
I'm an undergraduate student from India, exploring the field of Scientific Machine Learning — where machine learning meets scientific computing. My interests lie in applying Python-based tools to problems like solving differential equations, accelerating fluid simulations, and enhancing noisy or low-resolution data, such as in medical imaging.
Currently, I'm pursuing my studies at IISER Trivandrum and working at the Center for Scientific Computing and Computational Mechanics at IIT Delhi, where I’m contributing to operator learning–based models, which have shown great promise in scientific simulations.
This is my first conference talk, where I hope to share what I'm learning and connect with the Python community driving much of this progress.
Session
From predicting weather and modeling fluids to optimizing financial markets, traditional simulations rely on solving partial differential equations (PDEs) or using data-driven machine-learning models. However, PDE solvers are often computationally expensive, and pure data-driven approaches struggle with limited or noisy data. Physics-Informed Machine Learning (PI-ML) offers a powerful alternative by embedding physics directly into deep learning models—combining the strengths of both worlds.
This talk will introduce Physics-Informed Neural Networks (PINNs) and extend beyond them to more advanced approaches like Neural Operators. We’ll explore how these techniques are transforming real-world applications—from fluid simulations in engineering to climate forecasting and even economic modeling.
Attendees will learn:
- How physics can enhance ML models for better generalization and efficiency.
- Practical implementation using Python frameworks like PyTorch, Deep-XDE, and NVIDIA PhysicsNeMo.
- Case studies where PI-ML models outperform traditional methods.
No deep math or PDE knowledge is required—this session is designed to be insightful, engaging, and accessible to ML practitioners, engineers, and researchers curious about bridging physics and AI.