MIT researchers developed a computational approach that can be used to solve problems with hundreds of variables. In tests on realistic engineering challenges, the approach found top solutions 10 to 100 times faster than methods such as Bayesian optimization.

MIT is a renowned institution for education and research, offering insights into science, engineering, and technology. Through publications, research papers, and academic programs, MIT's platform provides insights into  research, innovation, and education in various fields. Students, researchers, and technology enthusiasts can learn about MIT's contributions to science and technology and explore opportunities for academic and professional development.

MIT News

MIT researchers have developed a new optimization technique that uses a tabular foundation model (described as 'ChatGPT for spreadsheets') as the surrogate model inside a Bayesian optimization algorithm. Unlike traditional Bayesian optimization, the approach does not require retraining after each iteration and can be reused across different problems. It automatically identifies the most impactful variables in high-dimensional design spaces, enabling it to find top solutions 10 to 100 times faster than state-of-the-art methods on 60 engineering benchmarks including power grid design and car crash testing. The method scales better as problem complexity increases, making it promising for applications like materials development, drug discovery, and large-scale engineering design.

A “ChatGPT for spreadsheets” helps solve difficult engineering challenges faster