ECE/SIOC 228 Machine Learning for Physical Applications
Instructor
Yuanyuan Shi, Assistant Professor, yyshi@eng.ucsd.edu
Office hours: Thursday 4-5 pm [Jacobs Hall Room 4401]
Teaching Assistants
Tawaana Homavazir, Office Hours: 2:00 pm - 3:00 pm Monday
Rishabh Jangir, Office Hours: 3:30 pm - 4:30 pm Tuesday
Srinivas Rao Daru, Office Hours: 3:30 pm - 4:30 pm Wednesday
Rohin Garg, Office Hours: 9:30 am - 10:30 am Thursday
Course Time and Location
Tuesday/Thursday, 2-3:20 pm, Warren Lecture Hall 2001
Course Structure
Part 0: Introduction & Fundamentals
Lecture 1: Introduction and Course Logistics [Lecture Note]
Lecture 2: Supervised Learning Setup [Lecture Note]
Lecture 3: Neural Networks and Backpropagation [Lecture Note]
Lecture 4: Optimization & Regularization in Deep Learning [Lecture Note]
Part 1: Modeling and Learning in Physical Systems
Lecture 5: Convolutional Neural Networks [Lecture Note]
Lecture 6-7: Recurrent Neural Networks [Lecture Note]
Lecture 8-9: Graph Neural Networks [Lecture Note]
Lecture 10: Neural Operators (Guest Lecture: Zongyi Li) [Lecture Note]
Lecture 11: Physics Informed Neural Networks [Lecture Note]
Review of Modeling and Learning [Lecture Note]
Part 2: Decision Making in Physical Systems
Lecture 12-13: Markov Decision Process [Lecture Note]
Lecture 14: Deep Reinforcement Learning I: Policy Gradient [Lecture Note]
Lecture 15: Deep Reinforcement Learning II: Policy Gradient with Variance Reduction [Lecture Note]
Lecture 18: Deep Reinforcement Learning III: Actor-Critic Method [Lecture Note]
Lecture 19: Deep Reinforcement Learning IV: Value Function Methods and DQN [Lecture Note]
Part 3: Guest Lectures (ML for various physical applications) and Projects
Lecture 16: Machine Learning for Power System: Route to Sustainability (Dr. Yize Chen) [Lecture Note ]
Lecture 17: Machine Learning for Robotics (Dr. Guanya Shi) [Lecture Note]
Lecture 20: Final Project Poster Presentation
Many thanks for the fun projects! Here are some example final projects presented at the poster event in Spring 2022!
Predicting Wildfire in California
Crown of Thorns Starfish Detection
Hurricane Trajectory Detection using LSTM
Machine Learning for Breast Cancer Diagnosis
Textbook and References
There is no single text that provides adequate coverage of the concepts and the background material in to serve the needs of this course. There is no required textbook for the course. We will use my lecture notes and the following other references/lecture notes for material, each of which has a part of the relevant material for the course.
Modeling and Learning in Physical Systems
Deep Learning, by Ian Goodfellow and Yoshua Bengio, and Aaron Courville
CS230 Deep Learning, Stanford, by Andrew Ng
Understanding Convolutions on Graphs, by Ameya Daigavane, Balaraman Ravindran, and Gaurav Aggarwal
Physics-Informed Neural Networks: A Deep Learning Framework for Solving Forward and Inverse Problems Involving Nonlinear Partial Differential Equations, by M. Raissi, P. Perdikaris, and G.E. Karniadakis
Decision Making in Physical Systems
An Introduction to Reinforcement Learning, by Richard Sutton
Reinforcement Learning, by David Silver
Deep Reinforcement Learning, by Sergey Levine