Session: DAC-02-02: Artificial Intelligence and Machine Learning for Challenging Real-World Problems in Design Automation
Paper Number: 117216
117216 - Counterfactuals for Design: A Model-Agnostic Method for Design Recommendations
We introduce Multi-Objective Counterfactuals for Design (MCD), a novel method for counterfactual optimization in design problems. Counterfactuals are hypothetical situations that can lead to a different decision or choice, and in this paper, the authors frame the counterfactual search problem as a design recommendation tool that can help identify modifications to a design, leading to better functional performance. MCD improves upon existing counterfactual explanation methods by supporting multi-objective queries, which are crucial in design problems, and by decoupling the counterfactual search and sampling processes, thus enhancing efficiency and facilitating objective tradeoff visualization. The paper provides a demonstration of MCD's core functionality using a two-dimensional test case, followed by three case studies of bicycle design that showcase MCD's effectiveness in real-world design problems. In the first case study, MCD is shown to excel at recommending modifications to query designs that can significantly enhance functional performance, such as weight savings and improvements to the structural safety factor. The second case study demonstrates that MCD can work together with a pre-trained language model to effectively suggest design changes based on a subjective text prompt. Lastly, the authors task MCD with increasing a query design's similarity to a target image and text prompt while simultaneously reducing weight and improving structural performance, demonstrating MCD's performance on a complex multimodal query. Overall, the method has the potential to provide valuable recommendations for design practitioners looking to optimize their designs and design automation researchers looking to interact intuitively with their models.
Presenting Author: Faez Ahmed Massachusetts Institute of Technology
Presenting Author Biography: Faez Ahmed is the d'Arbeloff career development assistant professor in the Department of Mechanical Engineering at MIT, where he leads the Design Computation and Digital Engineering (DeCoDE) lab. His research focuses on developing new machine learning and optimization methods to study complex engineering design problems. Before joining MIT, Ahmed was a postdoctoral fellow at Northwestern University and completed his Ph.D. in mechanical engineering at the University of Maryland. He also worked in the railway and mining industry in Australia, where he pioneered data-driven predictive maintenance and renewal planning efforts.
Authors:
Lyle Regenwetter Massachussetts Institute of TechnologyYazan Abu Obaideh ProgressSoft
Faez Ahmed Massachusetts Institute of Technology
Counterfactuals for Design: A Model-Agnostic Method for Design Recommendations
Paper Type
Technical Paper Publication