What is it about?
This paper aims to address the unsolved problem of finding truly innovative designs. This is approached by teaching an estimator to directly predict the set of all 'feasible' designs. The estimator is taught using a Surrogate Based Optimization (SBO) problem that is modified to use a classification surrogate and reinforcement learning infill. This modification leads to the discovery of the first ever application of classification surrogates in design optimization. It is envisioned that a practitioner using this approach would find a set of radical feasible designs large enough to survive the uncertainty of all stages of design resulting in an innovative solution.
Featured Image
Photo by Yerlin Matu on Unsplash
Why is it important?
This work is important because the classification surrogate is a massive family of tools that have been untouched for use in design optimization problems.
Perspectives
This work is born out of the frustration of real life design engineering, where the design point often changes after optimization. The investigation into classification surrogates is an attempt to try and uncover new ways of learning design spaces that compliment the real life abuse of optimization.
Said Mouhaiche
The University of Sydney
Read the Original
This page is a summary of: Lazily Reformulating Design Optimization as a Classification Problem, January 2025, American Institute of Aeronautics and Astronautics (AIAA),
DOI: 10.2514/6.2025-0112.
You can read the full text:
Resources
Contributors
The following have contributed to this page
