Welcome to the course Machine Learning and the Physical World. The course is focused on machine learning systems that interact directly with the real world. Building artificial systems that interact with the physical world have significantly different challenges compared to the purely digital domain. In the real world data is scarce, often uncertain and decisions can have costly and irreversible consequences. However, we also have the benefit of centuries of scientific knowledge that we can draw from. This module will provide the methodological background to machine learning applied in this scenario. We will study how we can build models with a principled treatment of uncertainty, allowing us to leverage prior knowledge and provide decisions that can be interrogated.
Lecturoes
| Lecture | Date | Lecturer |
|---|---|---|
| Introduction | che29 | |
| Simulation | Nicolas | |
| Quantification of Beliefs | che29 | |
| Gaussian Processes | che29 | |
| Practical Gaussian Processes | che29 | |
| Emulation | che29 | |
| Sequential Decision Making Under Uncertainty | che29 | |
| Probabilistic Numeric | che29 | |
| Reinforcement Learning | che29 | |
| Experimental Design | Nicolas | |
| Sensitivity Analysis | Nicolas | |
| Multifidelity Modelling | che29 | |
| Stirred Tank Reactor Design | Bethany Conroy | |
| Generative Models | che29 | |
| TBD | che29 | |
| TBD | che29 |
Worksheets
The worksheets below include a more detailed description of the material that we have gone through in the lectures. The aim is for the lectures to set the scene while these worksheets should clarify the details and make the material concrete.
| Lecture | Date | Lecturer |
|---|---|---|
| Quantification of Beliefs | che29 | |
| Gaussian Processes | che29 | |
| Practical Gaussian Processes | che29 | |
| Sequential Decision Making Under Uncertainty | che29 |
Assignments
The work should be submitted on the moodle page.
| Assignment | Date | Lecturer |
|---|---|---|
| Gaussian Processes | che29 | |
| Sequential Decision Making | che29 |
Project
The final part of the unit is a group project. You should do this in groups of 3.
Report
You will submit a report on the moodle page of the course. The format of this should be as a paper. Use the style files and instructions here https://nips.cc/Conferences/2023/PaperInformation/StyleFiles. This means that the final submission should be 9 pages of content which should be plenty for you to describe the work and conclusions. Please do not add an Appendix to the report instead think about what you want to say carefully. Remember, you do not need to explain for example Gaussian process basics if this is something that you use for your project instead spend time explain what you have done in terms of the specifics of your project and if any novelties (say for example interesting kernels) that you have used. Try to focus on the narrative, motivate why you are making specific choices to address the problem at hand.
Viva
As the final part of the assessment you will have a short oral examination. This presentation and viva is on your own material so do not think of it as a test, it is for you to make sure that we have understood what you have done so that all is not on the report. You will prepare a 5-10 minute introduction to what you have done and then we will discuss so that we make sure we gotten everything. As soon as we have gone through the reports we will send out a link where you can book a presentation/viva time. The viva will take less than 25 minutes in total and it will also be a place for you to get feedback on the work and give us feedback on the course to tell us what we did well and what we can improve upon.
Individual
While this is a group project we are also required by the department to collect information on what each individual have done for the project. To do so each group member should submit a up to one page document with the following content.
- the CSRid of all members in your group
- summary of your personal contribution to the work of the team
- your assessment of the contributions made by each other member of the team
Final Submission
One member of the team submits the final report and each member of the team submits their document outlining the contributions. Just to reduce my detective work it would be great if you put the CSRid of all members of the group in the report and on your individual submission.
People
Professor Neil D. Lawrence
Professor Carl Henrik Ek
Dr. Nicola Durrande