From data to model: how to build up a biomechanical mathematical model starting from recorded data of muscular activity.
Simona Galliani, Harnoor Saini, Johannes Walter
Institute for Modelling and Simulation of Biomechanical Systems
In this seminar we focus on data-driven biomechanical modeling and promote the interdisciplinary skills of the PhD students. We do so with a case study which starts from gathering data on muscular contraction through EMG (Electromyography) and MoCap (Motion Capture) to integrate these data into a biomechanical model.
9:45 on Tuesday 18 June (startup session)
SimTech Seminar Room 0.015 (PWR 5a)
Presence, Regular and active participation in discussions, Assignments
If you are interested, please contact firstname.lastname@example.org such that the number of participants can be anticipated.
Stochastic and Statistical Topics in Modeling and Simulation: Artificial Neural Networks from a statistical point of view
Prof. Dr.-Ing. Wolfgang Nowak
Artificial neuronal networks (ANNs) are a powerful concept for machine learning, i.e., to identify and represent functional relations that are hidden within data sets. However, classic ANNs start as fully stupid before training, can only learn what is inside the given data, and are so in-transparent that humans do not learn anything. Hence, we need to build a proper and deep and theory-underpinned understanding of several topics:
- Fundamentals of ANNs
- Physics-constrained ANNs
- Types & choosing structures of ANNs
- Proper testing and validation
- Choosing the training data
- Inference instead of optimization
- The training problem and its algorithms
- Bayesian instead of regularization
- Overfitting/underfitting, Regularization
- Uncertainty and stochasticity
The presented talks will include theory, corresponding algorithms and, whenever possible, instructive programming demos. Participants are invited from science, engineering, mathematics, statistics and computer science.
Mondays, 11:30, starting on 29. April 2019 (startup session)
Room 0.015 in SimTech (PWR 5a)
Tutorial talks by participants (30+min), active in-class discussion
If you are interested, please contact me under email@example.com so that the number of participants can be anticipated.