Duration
10h Th, 20h Pr
Number of credits
| Master MSc. in Data Science, professional focus | 5 crédits | |||
| Master MSc. in Data Science and Engineering, professional focus | 5 crédits |
Lecturer
Language(s) of instruction
English language
Organisation and examination
Teaching in the second semester
Schedule
Units courses prerequisite and corequisite
Prerequisite or corequisite units are presented within each program
Learning unit contents
This course explores the application of machine learning to gravitational-wave astronomy. Through lectures and practical work, students investigate how machine learning methods can identify and characterise astrophysical signals in noisy interferometer data.
The following list indicates the range of applications and methods from which course topics will be selected. The selection and depth of coverage may vary according to students' prior knowledge, course progression, available time and developments in the field. Not all listed topics will necessarily be covered.
Potential topics include:
- Gravitational-wave detection: identifying signals from astrophysical sources, including binary black holes and neutron stars.
- Signal processing and denoising: exploring methods such as sparse dictionary learning and autoencoders for denoising and glitch mitigation.
- Time-series analysis: applying recurrent neural networks, including long short-term memory (LSTM) networks, to transient signal detection.
- Convolutional neural networks (CNNs): analysing gravitational-wave data and time-frequency representations for detection and classification.
- Anomaly detection: identifying unusual features in interferometer data, including applications of convolutional autoencoders.
- Generative adversarial networks (GANs): exploring synthetic data generation and data augmentation for machine learning applications.
- Gradient boosting, including XGBoost: classification and regression applications in gravitational-wave data analysis.
Learning outcomes of the learning unit
By the end of the course, students should be able to:
- Formulate a gravitational-wave data analysis problem as a machine learning task.
- Prepare and analyse interferometer datasets, recognising the challenges posed by noise, glitches and limitations in the data.
- Implement selected machine learning methods using Python and Jupyter notebooks.
- Evaluate model performance and interpret the results in their astrophysical context.
- Explain and justify methodological choices, assumptions and limitations.
- Critically assess scientific explanations, code and results, including LLM-generated material where used.
Prerequisite knowledge and skills
Students should have a solid foundation in Python programming and basic machine learning concepts. Prior knowledge of gravitational-wave astronomy is also recommended.
Planned learning activities and teaching methods
The course combines lectures with practical projects using real gravitational-wave datasets and Jupyter notebooks. Students develop and apply machine learning methods under the instructor's guidance. Group discussions and student presentations provide opportunities to explain methodological choices, compare approaches and interpret results.
During this academic year, the instructor will pilot the use of large language models (LLMs), with the aim of providing each student with personalised lecture slides and notebooks based on the instructor's course materials. Personalisation will draw on an initial questionnaire and feedback during the course. The approach will be reviewed throughout the year and may be adapted or discontinued in response to student feedback, its educational effectiveness and the availability of technical resources.
Learning activities may also include guided critical review of scientific explanations, code and results, including LLM-generated material. Students may use LLMs to support these review activities, following guidance and prompts provided by the instructor. Discussions will focus on assessing accuracy, examining assumptions and limitations, and justifying proposed corrections.
Mode of delivery (face to face, distance learning, hybrid learning)
Face-to-face course
Further information:
The course will be delivered in person, with flexibility for online discussions or assignments as needed.
Course materials and recommended or required readings
Platform(s) used for course materials:
- MyULiège
Further information:
Course materials will be made available through MyULiège and will include lecture slides, Jupyter notebooks, Python tutorials and selected research papers on machine learning applied to gravitational-wave data analysis. Any required readings will be identified explicitly.
Personalised versions of the lecture slides and notebooks may also be provided as part of the LLM teaching pilot. LLM-adapted materials will be identified as such and may be revised during the course following review and student feedback.
Exam(s) in session
Any session
- In-person
oral exam
Written work / report
Further information:
- An oral exam to assess students' understanding of both gravitational-wave concepts and the machine learning techniques discussed during the course.
- A homework assignment that will be graded, focused on applying machine learning methods to gravitational-wave data.
Work placement(s)
Organisational remarks and main changes to the course
Sessions will typically combine discussion of the relevant gravitational-wave and machine learning concepts with practical work using Jupyter notebooks.
The main teaching development this academic year is the pilot of personalised learning materials described above. Students will be invited to provide feedback on their clarity, usefulness and accuracy throughout the course. This feedback will inform revisions to the materials and decisions about the scope and continuation of the pilot. Any adjustments will be communicated to students during the course.
Contacts
Maxime Fays
(maxime.fays@uliege.be)
Room 4.43 Bât. B5A
Inter. fondamentales en physique et astrophysique (IFPA)
Quartier Agora allée du six Août 19
4000 Liège
Téléphone de service: +32 4 3663643