ECE228 and SIO209 Machine learning for physical applications, Spring 2019

Professor Peter Gerstoft, Spiess Hall 462, Gerstoft@ucsd.edu
TA Siva Prasad Varma Chiluvuri, sivapvarma@gmail.com
TA Harshuk Gupta, h6gupta@eng.ucsd.edu
TA Ruixian Liu, rul188@eng.ucsd.edu
Location: SOLIS 107
Time: Monday and Wednesday 5-6:20pm


Sylabus: Machine learning has received enormous interest. To learn from data we use probability theory, which has been the mainstay of statistics and engineering for centuries. The class will focus on implementations for physical problems. Topics: Gaussian probabilities, linear models for regression, linear models for classification, neural networks, kernel methods, support vector machines, graphical models, mixture models, sampling methods, sequential estimation. Prerequisites: graduate standing.

It is not a computer science class, so we go slowly through the fundamentals to appreciate the methods, implement these. We will discuss their use in Physical Sciences. In the first part of the class we focus on theory and implementations. We will then transition to focus on machine learning final projects.


Books (all available online): Main book: Chris M Bishop Pattern Recognition and Machine Learning . A third party Matlab implementation of many of the algorithms in the book.
Other good books:
Hastie and Tibshirani The Elements of Statistical Learning
Kevin P. Murphy: Machine Learning: A Probabilistic Perspective. UCSD license . Matlab codes used in Murphy's book.

Online resources: While not required, I recommend taking these classes. Both are online classes are excellent.
Statistical Learning by Hastie and Tibshirani. My favorite class.
Andrew Ng's Coursera class, Machine learning. This was the first class offered by Coursera.

Grading: Full scale of the letter grade. Grade consists of About 50 % homework and 50% final-project (10% poster, 10 % code and 30 % Report). Your and my purpose is to lean, so a good effort is sufficient.
Homework Cody homework will be graded. I'm a strong believer in leaning by doing. Thus we will have computer-based homework each week.


Final project: Propose a topic before May 1. We will make teams on ~April 26. Propose a topic before ~May 5. Preliminary presentation 20 May, Final Poster 5 June. Report due Saturday June 15.
Suggested topics (see last years projects Spring 2018 and Spring 2017 ):

Schedule (15 Lectures. Two presentation classes )

Hastie CH 9
Lecture Date Topic Required reading Assignments
1April 1 Introduction, probability theory [Slides Annotated Slides ]
Bishop CH 1.0-1.2
2April 3 Gaussian probability theory [Slides, Annotated Slides ]
Question to lecture 2 in Stanford CNN lecture
Bishop CH 1.5, 1.6, 2.3
3April 8 Gaussian probability theory, Gaussian processes [Slides Annotated Slides]
Question to lecture 3-4 in Stanford CNN lecture
Bishop CH 2.3, 6.4
Murphy ch15
Homework 1 due 4/22
4April 10 Gaussian Processes, Linear models for Regression
Diego Nozal: Using Gaussian Processes for outdoor Concert sound fields [Slides Annotated Slides]
Question to lecture 5 in Stanford CNN lecture
Bishop CH 3.0-3.1 Homework 2 due 4/22
5April 15 Linear models for Regression [Slides Annotated Slides]
Question to lecture 6 in Stanford CNN lecture
Bishop CH 3.2-3.3
6April 17
Question to lecture 7 in Stanford CNN lecture
Emma Ozanich: ML on NOAA climate data using jupyter and GPU [Slides Annotated Slides]
GPU Jupiter homework relased
due 5/10
7April 22 Question to lecture 8 in Stanford CNN lecture
Linear models for Classification
Backpropagation[Slides Annotated Slides]
Bishop 4.0-4.3.2, 4.3.4, 5.0-5.2 Homework 3 (NN) due 5/5.
Subset of MNIST Data to use for HW3 on you laptop
8April 24 Question to lecture 9 in Stanford CNN lecture

Kernel methods [Slides Annotated Slides]
Bishop CH
9April 29 Question to lecture 10 in Stanford CNN lecture
Christopher Johnson: Machine learning with CNN in seismology.
[Slides Annotated Slides]
Bishop CH 6
10May 1 Question to lecture 11 in Stanford CNN lecture
Support vector machines [Slides Annotated Slides]
Bishop CH 7
11May 6 Question to lecture 12 in Stanford CNN lecture
K-means, EM and Mixture models [Slides Annotated Slides]
Bishop CH 9 Project proposals due
Homework 5(SVM +Dictionary Learning) due 5/29
12May 8 Question to lecture 13 in Stanford CNN lecture
Mike Bianco: Dictionary leaning
Read Bianco's Dictionary learning paper
-May 13 No Class, work on Final project.
13May 15 TA Siva Varma: Jupyter homework and leftover material on Dictionary Leanring
14May 20 Question to lecture 13 and 14 in Stanford CNN lecture
Sequential estimation
[Slides Annotated Slides]
Bishop CH 13
1 May 22 mid-project presentation of final-projects
15May 29 Question to lecture 15 in Stanford CNN lecture
Sequential estimation and Sparse models
16June 3 Question to lecture 16 in Stanford CNN lecture
Sparse models

Read Chapter 2 for an introduction to sparse problems
Murphy 13.1, 13.3, 13.6.1
Homework 6 (Kalman Filter and Sparse signal recovery) due 6/10
2 June 5 Final project poster presentation, Atkinson Hall and Lobby, 5-8 pm Final report due Saturday 15 June.

Spring 2018 class

Spring 2017 class