Machine Learning for Physics and Astronomy (Honours)

3 EC

Semester 1, period 1

5093MLFP3Y

Owner Bachelor Natuur- en Sterrenkunde (joint degree)
Coordinator Edan Lerner
Part of Bachelor Physics and Astronomy (Joint Degree), year 3

Course manual 2023/2024

Course content

In recent years, Machine Learning tools have demonstrated their impressive potential to tackle a range of challenging problems related to classification, optimization, discrimination, inter- and extrapolation, efficient parameterizations, and pattern generalization, among several others. In this course, we will present an overview of the foundational concepts in Machine Learning and highlight some of their applications to various areas of Physics and Astronomy, from particle and astroparticle physics to condensed matter. The course will be a dynamical combination of theoretical concepts and formalism, discussion of practical applications, and hands-on tutorials by means of Python Jupyter notebooks.

Study materials

Literature

  • See course guide

Syllabus

  • See course guide

Practical training material

Objectives

  • Students become familiar with, and understand the differences between, various classes of machine learning algorithms
  • Students are able to apply a selection of machine learning tools to data

Teaching methods

  • Lecture
  • Presentation/symposium
  • Laptop seminar

Learning activities

Activity

Hours

Hoorcollege

14

Tentamen

3

Werkcollege

12

Self study

55

Total

84

(3 EC x 28 uur)

Attendance

Programme's requirements concerning attendance (TER-B):

  • Each student is expected to participate actively in each component of the programme that he/she signed up for. A student that does not attend the first two seminars of a course, will be administratively removed from the seminar group. A request for reregistration for the seminars can be applied to the programme coordinator.
  • If a student cannot attend an obligatory component of a programme's component due to circumstances beyond his control, he must report in writing to the relevant teacher as soon as possible. The teacher, if necessary after consulting the study adviser, may decide to issue the student a replacing assignment.
  • It is not allowed to miss obligatory commponents of the programme if there is no case of circumstances beyond one's control.
  • In case of participating qualitatively or quantitatively insufficiently, the examiner can expel a student from further participation in the programme's component or a part of that component. Conditions for sufficient participation are set down in advance in the course manual.
  • In addition to the above mentioned rules, in the first semester of the first year a student should be present in at least 80% of the seminars. Moreover, participation to midterm tests and obligatory homework is required. If the student does not comply with these obligations, the student is expelled from the resit of this course. Students in the double Bachelor's degree programme Mathematics and Physics are exempted from this requirement. In case of personal circumstances, as described in OER-A Article A-6.4, a different study plan will be made in consultation with the study advisor.

Additional requirements for this course:

n/a

Assessment

Item and weight Details

Final grade

0.6 (60%)

literature review project presentation

0.4 (40%)

homework assignments

1) homework assignments amount to 40% of the final grade; the rubric for the assignments is: 

  • nothing handed in: 0 points 
  • solution partially complete/correct: 1 point 
  • solution mostly complete/correct: 2 points 
  • fully complete and correct solution: 3 points 

Assignment should be completed individually, and submitted on the course’s canvas webpage. 

2) Literature review project presentation amounts to 60% of the final grade.  
Instructions: choose a ML4phys paper, review & explain it in a pedagogical manner in front of fellow students (in pairs) (20 minutes presentation, 10 minutes Q&A). Make sure to highlight: 

  • the physics problem that is being addressed 
  • the details of the machine learning algorithm adopted (and the motivation thereof) \
  • the connection of the paper with the course’s material/spirit/topics 
  • the main results, and the role played by the ML algorithm 
  • and how this approach improves over more traditional strategies 

Inspection of assessed work

via Canvas

Fraud and plagiarism

The 'Regulations governing fraud and plagiarism for UvA students' applies to this course. This will be monitored carefully. Upon suspicion of fraud or plagiarism the Examinations Board of the programme will be informed. For the 'Regulations governing fraud and plagiarism for UvA students' see: www.student.uva.nl

Course structure

Weeknummer lecture exercise
1

-Introduction 

-overview/review of ML applications to physics 

 -unsupervised ML: clustering, PCA 

Dimensional reduction & clustering 
2

-neural networks (NNs) 

-feed-forward & backpropagation 

-optimization 

-convolutional NN 

gradient descent 
3

unsupervised ML cont’d:  

-autoencoders 

-Generative Adversarial Networks 

-Restricted Boltzmann Machines 

RBM tutorial 
4

supervised ML: introduction 

-overfitting 

-bias-variance tradeoff 

tbd 
5

Guest lectures:  

- Alberto Pérez de Alba Ortı́z 

- Ryan van Mastrigt 

tbd 
6

supervised ML cont’d:  

-support vector machines 

-kernel methods 

Lecture: how are physicists contributing to understanding how ML algorithms work? 
7 student presentations  student presentations 

Timetable

The schedule for this course is published on DataNose.

Contact information

Coordinator

  • Edan Lerner