Physics-Informed Machine LearningModule description #41190 / Version 1
- 3 CP
- Portfolio examination
- Graded
- Winter and summer semester
- English
| Module / Version | #41190 / #1 |
|---|---|
| Validity | Since Winter semester 2024/25 |
| Available language(s) | English, German |
| Credit points | 3 |
| Person responsible for the module | Haufe, Stefan |
| Grading | Graded |
| Exam format | Portfolio examination |
| Language(s) of instruction | English |
Affiliation
| Faculty | Fakultät IV |
|---|---|
| Institute | Institut für Softwaretechnik und Theoretische Informatik |
| Academic chair | 34352700 FG S-Professur Unsicherheit, inverse Modellierung und maschinelles Lernen (UNIML) |
Contact
| Office | MAR 4-3 |
|---|---|
| Contact person | Huseynov, Ismail |
| E-mail address | huseynov@campus.tu-berlin.de |
| Website | http://tu.berlin/uniml |
Learning outcomes
Knowledge: Participants will have gained a comprehensive understanding of the core ideas behind physics-informed machine learning (PIML). PIML integrates traditional mathematical differential equation modeling, governing the behavior of many physical systems, with modern neural network based statistical learning algorithms, leading to enhanced efficiency and reliability in predictive modeling of physical systems. Students will have obtained an extensive overview of the manifold ways in which physical concepts are encoded into modern machine learning architectures, with individual deep insights into a topic of choice.
Skills: Students will have acquired or refined skills to independently review and systematically structure scientific literature and will have gained experience in presenting their literature research to a critical audience as well as in participating in scientific discussions.
Competencies: Students will be able to critically reflect on novel scientific results in the scope of physics-informed machine learning. They are able to relate these results to the literature and assess their quality and impact based on adequate metrics.
Skills: Students will have acquired or refined skills to independently review and systematically structure scientific literature and will have gained experience in presenting their literature research to a critical audience as well as in participating in scientific discussions.
Competencies: Students will be able to critically reflect on novel scientific results in the scope of physics-informed machine learning. They are able to relate these results to the literature and assess their quality and impact based on adequate metrics.
Content
Physics-informed machine learning (PIML) represents a novel approach that merges established physical principles, such as ordinary and partial differential equations with deterministic and stochastic settings, with the capabilities of machine learning algorithms. By embedding physical laws into the learning framework, PIML enables more precise and robust predictions, even within complex, uncertain, and high-dimensional domains. This integration of noisy data and mathematical physics models is typically realized through neural networks-based or kernel-based regression models, offering efficient and straightforward implementations for forward and inverse problems.
In this seminar, participants will analyze the core principles of PIML, focusing on the numerical analysis of forward and inverse problems prescribed by a class of ordinary differential equations (ODEs), partial differential equations (PDEs), fractional-order differential equations (FDEs), delay differential equations or time-delay systems (DDEs), and stochastic differential equations (SDEs) based on existing literature. We will explore the applications of PIML across diverse fields (e.g., fluid dynamics, medical imaging, power systems, etc.) governed by various mathematical modeling problems. Furthermore, in the homework task, participants will implement PIML algorithms based on various mathematical problems to enhance their theoretical knowledge with practical applications using Python and its various ML toolboxes.
Covered topics may include
- Physics-informed neural networks (PINNs)
- Forward-Backward Stochastic Neural Networks
- Fractional physics-informed Neural Networks (fPINNs)
- Uncertainty Quantification in PINNs
- Curse of Dimensionality problem in PINNs
- Neural Ordinary Differential Equations
- Neural Stochastic Differential Equations
- Universal Differential Equations
- Diffusion models
- Invariant and equivariant networks
- Spiking neural networks
- Applications in fluid dynamics and mechanics, power systems, computational neuroscience, systems biology, finance
In this seminar, participants will analyze the core principles of PIML, focusing on the numerical analysis of forward and inverse problems prescribed by a class of ordinary differential equations (ODEs), partial differential equations (PDEs), fractional-order differential equations (FDEs), delay differential equations or time-delay systems (DDEs), and stochastic differential equations (SDEs) based on existing literature. We will explore the applications of PIML across diverse fields (e.g., fluid dynamics, medical imaging, power systems, etc.) governed by various mathematical modeling problems. Furthermore, in the homework task, participants will implement PIML algorithms based on various mathematical problems to enhance their theoretical knowledge with practical applications using Python and its various ML toolboxes.
Covered topics may include
- Physics-informed neural networks (PINNs)
- Forward-Backward Stochastic Neural Networks
- Fractional physics-informed Neural Networks (fPINNs)
- Uncertainty Quantification in PINNs
- Curse of Dimensionality problem in PINNs
- Neural Ordinary Differential Equations
- Neural Stochastic Differential Equations
- Universal Differential Equations
- Diffusion models
- Invariant and equivariant networks
- Spiking neural networks
- Applications in fluid dynamics and mechanics, power systems, computational neuroscience, systems biology, finance
Module components
Compulsory area
The following courses are mandatory for this module:
| Courses | Type | Number | Frequency | Language | Hours per week | Course catalog |
|---|---|---|---|---|---|---|
| Physics-informed Machine Learning | Seminar | Winter and summer semester | [en] | 2 | ISIS |
Workload and credit points
Physics-informed Machine Learning (Seminar)
| Workload description | Multiplier | Hours | Total |
|---|---|---|---|
| Attendance | 15 | 2 h | 30 h |
| Pre/post processing | 15 | 4 h | 60 h |
| 90 h | |||
The workload of the module adds up to 90 hours. The module therefore comprises 3 credit points.
Description of teaching and learning methods
The module consists of a seminar, which will either be held in a weekly format or as a block seminar over two full days at the end of the semester.
At the beginning of the semester, students will choose a topic based on a provided collection of published material.
Students will become acquainted with the topic and prepare a presentation to be held in class, where they regularly confer with their assigned advisor.
The course grade is composed of the student's presentation and a homework comprising a programming task.
At the beginning of the semester, students will choose a topic based on a provided collection of published material.
Students will become acquainted with the topic and prepare a presentation to be held in class, where they regularly confer with their assigned advisor.
The course grade is composed of the student's presentation and a homework comprising a programming task.
Requirements for participation and examination
Desirable prerequisites for participation in the courses:
A BSc degree in Computer Science, Mathematics or Engineering is recommended. Successful completion of an introductory module on ML such as "Machine Learning I" or "Machine Intelligence I" is also recommended as is solid knowledge of mathematics (particularly partial differential equations).
Mandatory requirements for the module examination registration:
This module has no examination requirements.
Completion of the module
| Grading | Graded |
|---|---|
| Exam format | Portfolio examination |
| Type of portfolio examination | 100 points per element |
| Language | English |
Test elements
| Name | Points/Weight | Category | Duration/Extent |
|---|---|---|---|
| (Deliverable assessment) Homework | 50 | written | 6 pages |
| (Deliverable assessment) Presentation | 50 | oral | 30 min |
Exam description
The module grade is calculated based on
1. The quality of a paper presentation (50%).
2. The quality of a homework comprising a practical programming task (50%).
1. The quality of a paper presentation (50%).
2. The quality of a homework comprising a practical programming task (50%).
Duration of the module
The following number of semesters is estimated for taking and completing the module: 1 semester(s).
This module can be started in the following semesters: Winter and summer semester.
Maximum number of participants
The maximum capacity of participants is 16.
Registration procedures
Students can sign up for the course in MOSES.
Recommended reading, lecture notes
Lecture notes
unavailable
Electronic lecture notes
unavailable
Recommended literature
No recommended literature given
Assigned programs
This module version is used in the following programs:
| Program / program version | Semester(s) | Curriculum component |
|---|---|---|
| Computer Engineering Computer Engineering (M. Sc.) - StuPO 2015 | SS 2025 | Wahlpflicht Studiengebiet B › Medientechnik und Mensch-Maschine-Interaktion / Digital Media und Human-Computer Interaction |
| WS 2025/26 | Wahlpflicht Studiengebiet B › Kognitive Systeme / Cognitive Systems | |
| SS 2026 | Wahlpflicht Studiengebiet B › Medientechnik und Mensch-Maschine-Interaktion / Digital Media und Human-Computer Interaction | |
| WS 2026/27 | Wahlpflichtbereich › Studiengebiet B › Kognitive Systeme / Cognitive Systems | |
| Computer Science (Informatik) Computer Science (Informatik) (M. Sc.) - StuPO 2015 | SS 2025 | Wahlpflicht Studiengebiete Fak. IV › Medientechnik und Mensch-Maschine-Interaktion / Digital Media and Human-Computer Interaction |
| WS 2025/26 | Studiengebiete › Medientechnik und Mensch-Maschine-Interaktion / Digital Media und Human-Computer Interaction | |
| SS 2026 | Studiengebiete › Kognitive Systeme / Cognitive Systems | |
| WS 2026/27 | Wahlpflichtbereich › Studiengebiet › Kognitive Systeme / Cognitive Systems | |
| Electrical Engineering Electrical Engineering (M. Sc.) - StuPO 2015 | SS 2025 | Wahlpflicht Studiengebiete Fak. IV › Kognitive Systeme / Cognitive Systems |
| WS 2025/26 | Wahlpflicht Studiengebiete Fak. IV › Medientechnik und Mensch-Maschine-Interaktion / Digital Media and Human-Computer Interaction | |
| SS 2026 | Wahlpflicht Studiengebiete Fak. IV › Medientechnik und Mensch-Maschine-Interaktion / Digital Media and Human-Computer Interaction | |
| WS 2026/27 | Wahlpflichtbereich › Studiengebiete Fak. IV › Medientechnik und Mensch-Maschine-Interaktion / Digital Media and Human-Computer Interaction | |
| Industrial Engineering and Management Industrial Engineering and Management (M. Sc.) - StuPO 2026 | WS 2026/27 | Engineering Sciences / Specialization › AI Engineering › Cognitive Systems |
Miscellaneous
Not specified