Academia Sinica Logo

Abstract

Machine learning techniques has evolved beyond conventional classification and regression paradigms. The corresponding examples include:
- Variational Autoencoders and Generative Adversarial Networks for anomaly detection;
- Invertible Neural Networks as alternatives to Markov Chain Monte Carlo methods;
- Physics-informed neural networks for determining medium characteristics and solving partial differential equations;
- Unsupervised clustering and classification techniques;
I will start with foundational principles of neural networks and proceed to the discussion of the above mentioned techniques, with focus on physic-related applications.