Physics-Informed Machine Learning: Curating Training Data - Part 2

Explore the second stage of the machine learning process in this 36-minute video lecture focusing on collecting and curating training data to inform physics-based models. Learn about data augmentation techniques incorporating known symmetries, the importance of coordinate systems, and the differences between simulated and experimental data. Discover the balance between big data and diverse data, strategies for generalizing models with physics, and the challenges of expensive and biased data collection. Delve into handling rare events, small signals, and hidden variables in physics-informed machine learning. Gain insights into discovering governing equations and the concept of digital twins. Enhance your understanding of AI/ML applications in physics through this comprehensive exploration of data curation techniques.

Intro
Augmenting Data with Physics
Coordinates Matter!
Simulated vs Experimental Data
Big Data vs Diverse Data
Generalizing Models with Physics
Data is Expensive
Data is Biased
Rare Events
Small Signals
Galileo Dropped the Ball
Hidden Variables
Preview: Discovering Governing Equations
The Digital Twin
Outro