Deep Learning for Fundamental Sciences Using High-Performance Computing

Explore how deep learning revolutionizes fundamental sciences in this 27-minute conference talk from the O'Reilly AI Conference. Discover how exabytes of data from complex instruments are analyzed to uncover the universe's secrets. Learn about NERSC, the mission supercomputing center for US fundamental science, and its use of TensorFlow for novel methods and applications in high-performance computing. Delve into topics such as the ATLAS detector, deep learning in physics, classification and regression problems, simulation and generation challenges, and particle physics applications. Understand the role of extreme computing, MPI distributed training, and Jupiter notebooks in advancing scientific research. Gain insights into supersymmetry, its sensitivity, and why deep learning is the right approach for fundamental sciences. Join the speaker in exploring the intersection of artificial intelligence and cutting-edge scientific discovery.

Introduction
The fundamental science
What is NERC
The secrets of the universe
The ATLAS detector
Deep learning in physics
Classification
Results
Regression problem
Simulation problem
Generation problem
Particle physics problem
Extreme computing
MPI distributed training
Jupiter notebooks
Conclusion
Call to help
Supersymmetry
Supersymmetry sensitivity
Why is deep learning right