COURSE 1: Deep Learning in Computational Mechanics: an Introductory Course
Course description:
| Day and Time | Monday 17th of February 2025, full-day course |
| Place | Arts et Métiers – ENSAM (Paris Campus) |
| Total teaching hours | Full-day course |
| Attendance Fee | 250 euros |
| Maximum Attendance | 90 Participants |
After attending this course, participants will have an overview over the prominent deep learning methodologies in computational mechanics. By applying neural networks to various problems from mechanics (ranging from forward simulations, topology optimization, and inverse problems to material modeling and anomaly detection), the participants learn how neural networks can be applied to engineering problems. The covered theory is supported by hands-on exercises (using Python), enabling the participants to solve the engineering problems with neural networks. Moreover, these exercise implementations may easily be adapted for the participant’s own purposes, providing a jump-start into the world of deep learning in computational mechanics.
The course follows the content of the second edition of the book "Deep Learning in Computational Mechanics: An Introductory Course" (978-3-030-76587-3 978-3-030-76586- 6), whose release coincides with the DTE & AICOMAS conference.
Objectives and target groups:
The course is aimed at engineers with an interest in deep learning for engineering applications. As the course is didactically designed to be easy to follow, no prerequisite knowledge is required. Basic Python knowledge is, however, advantageous for the exercises. Although it is an introductory course, the focus will be on novel methods at the state-of-the- art, making the course relevant not only for beginners but also for users with more extensive knowledge of deep learning
Course Outline:
| Time | Session |
|---|---|
| 9:00-10:30 | Introduction to Deep Learning in Computational Mechanics |
| 11:00-12:30 | Inverse Problems |
| 13:30-15:00 | Generative Approaches |
| 15:30-17:00 | Surrogate Modeling |
Scientific Areas Covered:
- Broad methodological overview of the state-of-the-art of deep learning in computational mechanics (covering the bulk of the scientific literature)
- Data-driven surrogate modeling (using PyTorch & TensorBoard)
- Physics-informed & physics-aware learning (to reduce the required amount of data)
- Inverse problems, topology optimization, anomaly detection & material modeling (as engineering applications)
Preliminary Time Schedule
The course will be a full-day course consisting of two 90-minute slots.
Lecturers:
