Splines vs. Neural Networks: How Novel Machine Learning Approaches Influence Design Optimization

Stefanie Elgeti

TU Wien, Vienna University of Technology

Product innovation is a multi-step process: a creative phase where ideas are born, an evaluation phase where the ideas are evaluated, and an implementation phase where these ideas become tangible. While computer-based assistance systems are already available for the latter two phases, creativity is often still considered an exclusively human attribute. However, recent advances in artificial intelligence (AI) have challenged this notion, as creative AI agents are increasingly integrated into our daily lives and have demonstrated their potential to create original content (e.g., ChatGPT, DALL-E, MuseNet, DeepDream). In light of these advances, a new field of research has emerged in the area of AI-enabled design processes, leading to a more-than-human design process in which a computer agent collaborates with a design team to efficiently and creatively explore the entire design space in search of novel design solutions.

To this end, we will demonstrate new technologies, such as how Variational Autoencoders (VAE) can be used to learn low-dimensional, yet feature-rich shape representations. This approach promises significant improvements in both performance and variety of shapes that can be learned. The resulting geometric representation is then incorporated into a shape optimization framework. In addition, we explore the potential of reinforcement learning (RL) as an optimization strategy. RL is based on the trial-and-error interaction of an agent with its environment. As such, RL can be characterized as experience-driven, autonomous learning. While not necessarily superior to classical optimization algorithms (such as gradient-based approaches) for a single optimization problem, based on the existing literature, we expect RL techniques to thrive when recurrent optimization tasks arise.

Biography

Stefanie Elgeti is an engaged member of the computational science and engineering community, contributing not only to research in the field of numerical analysis and design, but also through active participation in committees and journal work. She holds the position of full professor at TU Wien, the Vienna University of Technology, in Austria, where she leads the research area of lightweight design. In her research, she combines numerical analysis of components and their manufacturing processes – specifically space-time finite element methods and isogeometric analysis – with numerical design – particularly shape and topology optimization. Recently, she has grown interested in mixed-initiative design approaches, where a human interacts with a generative AI to produce novel designs. Stefanie Elgeti formerly chaired the ECCOMAS Young Investigators Committee and co-chaired the GAMM Juniors. She serves as a member of the editorial board for both the International Journal Engineering with Computers (EWCO) and Advances in Computational Science and Engineering (ACSE). Additionally, she heads the scientific advisory board of the Austrian Center for Digital Production. In 2020, she was the recipient of the ECCOMAS Olgierd Cecil Zienkiewicz Young Investigators award.