Lecture
A glimpse into my research between Bayesian Optimization and Mechanics
- Date
- Monday 24 July 2023
- Time
- Location
-
Five years to understand that... - A glimpse into my research between Bayesian Optimization and Mechanics
Abstract:
In this talk, I will give an overview of my research over the last five years, which has mainly focused on the application of Bayesian optimization (BO) techniques to solve high-dimensional problems.
An important aspect of my research has been the development of novel algorithms that use both linear and nonlinear mappings to transform high-dimensional spaces into lower-dimensional ones. These mappings facilitate exploration and exploitation of the optimization landscape and enable more efficient and effective search strategies for small optimization budgets. A major area of application for the methods I have developed is the field of structural mechanics. I have successfully applied these techniques to address problems such as topology optimization of mechanical structures subjected to static and crash loads, material characterization for composite materials, and parameter optimization for automotive components. I will provide specific examples of how the combination of BO and structural mechanics has led to significant improvements in design optimization and performance. In addition, I have recently focused on meta-learning (dynamic configuration of algorithms in particular), exploratory landscape analysis, and benchmarking. I will highlight these advances in more detail and emphasize their importance in building flexible and scalable frameworks for efficient problem solving.
In conclusion, my aim is to present my findings and ongoing work as a motivation to invest resources in fostering collaboration between AI and engineering design. Such collaboration will be mutually beneficial for both fields and lead to building an interdisciplinary and dynamic community that can have a broader societal impact.
