date: Wednesday 07.02.2025
time: TBA
venue: Campus RJ - Room 1.201
participation-link: TBA
recording-link: TBA
| TITLE | SPEAKER | ABSTRACT | INVITED | CONFIRMED | |
|---|---|---|---|---|---|
| 0 | Physics informed machine learning | Christoph Würsch | True | False | |
| 1 | Introduction: AI/ML Activites at OST | JuanPi Carbajal | True | True | |
| 2 | From ML to AI | JuanPi Carbajal | True | True | |
| 3 | Causaly informed machine learning | Martin Surner | Applications relying on machine learning and statistical learning techniques, such as neural networks, have significantly grown over the past decade. Nevertheless, as these techniques learn only from observational data, they suffer from spurious correlations that may limit their performance in domain shifts. We will discuss an approach that guides modeles during the training phase using user-defined causal knowledge. Such guidance reduces the risk of spurious correlations, allowing for a focus on the causal structure. | True | False |
date: Thursday 12.09.2024
time: 10:30
venue: Campus RJ - Room 1.201
participation-link: None
recording-link: None
| TITLE | SPEAKER | ABSTRACT | INVITED | CONFIRMED | |
|---|---|---|---|---|---|
| 0 | Introduction to the seminars and overview of SciML | JuanPi Carbajal | I will introduce the seminars IET-KI-Treffen and give and overview on the current developments of Scientific Machine Learning at IET. | True | True |
| 1 | EFT-Game | Simon Walser | A game developed together with the physiotherapists. It's basically a training apparatus currently used by the Hockey players of the Rapperswil Jona Lakers to train their executive functions. Here you can find an SRF article about the project. | True | True |
| 2 | Agent-Based Model to Assess Non-Technical Challenges to Monitoring Combined System Overflows (waste water) | Samuel Derwort | This talk focuses on the evaluation of policies to improve sensor adoption in combined sewer overflow (CSO) tanks in Switzerland. A prototype of an agent-based model (ABM) is developed based on psychological and social simulation theories and context knowledge. Data sets of different categories are combined and integrated into the prototype, and three scenarios are developed evaluating the impact of policies: the planning instrument GEP, a regulation requiring sensor installation in CSO tanks, and better monitoring technology. Results suggest that policy ii) is most effective, with most agents adopting sensors shortly after the start of the simulation period in 2017. However, absolute results are not reliable and relative results must be carefully examined. The limited availability of empirical data, such as historical data of sensor adoption, leads to great uncertainties, and model parameters are adjusted ad hoc. The multidisciplinary approach of ABM can be useful for system understanding and opens up new paths for research. Further work should concentrate on data collection, refinement of the agent behavior model using the Theory of Planned Behavior, and the opinion dynamics model using survey data. | True | True |
| 3 | Inferring emergent dynamics with equation-free modeling | Reza Housseini | Recent developments in multiscale computation allow the solution of coarse equations for the expected macroscopic behavior of microscopically evolving particles without ever obtaining these coarse equations in closed form. The closure is obtained on demand through appropriately initialized bursts of microscopic simulation. The effective coupling of microscopic simulators with macrosocopic behavior requires certain decisions about the nature of the unavailable coarse equation. Such decisions include (a) the highest spatial derivative active in the coarse equation, (b) whether the equation satisfies certain conservation laws, and (c) whether the coarse dynamics is Hamiltonian. These decisions affect the number and type of boundary conditions as well as the algorithms employed in good solution practice. This iformation is however not directly available. In this talk we will showcase the application of an inference algorithm to discover these hidden properties, in a simulation of multiphase bubbly flow. | True | True |