Symbolic Regression for Scientific Discovery

Summary

Symbolic regression is an approach to explain observed data, using symbolic expressions. Compared to deep-learning-based black-box methods, those expressions are relatively human understandable, which is critical especially when non-machine-learning (ML) experts apply ML to problems in their research domains such as physics, chemistry, and material science. We discuss the potential of symbolic regression for scientific discovery (SRSD) applications, proposing new SRSD datasets and evaluation metrics.

Related Publications

Rethinking Symbolic Regression Datasets and Benchmarks for Scientific Discovery

Journal of Data-centric Machine Learning Research (DMLR)

Yoshitomo Matsubara, Naoya Chiba, Ryo Igarashi, Yoshitaka Ushiku

This work was mainly done while I was a research intern at OMRON SINIC X Corporation.

OpenReview Video Preprint Code SRSD-Feynman (Easy) SRSD-Feynman (Medium) SRSD-Feynman (Hard) SRSD-Feynman (Easy w/ Dummy Vars.) SRSD-Feynman (Medium w/ Dummy Vars.) SRSD-Feynman (Hard w/ Dummy Vars.)

A Transformer Model for Symbolic Regression towards Scientific Discovery

NeurIPS 2023 AI for Science Workshop (Oral)

Florian Lalande, Yoshitomo Matsubara, Naoya Chiba, Tatsunori Taniai, Ryo Igarashi, Yoshitaka Ushiku

This work was mainly done while I was a research intern at OMRON SINIC X Corporation.

OpenReview Code

SRSD: Rethinking Datasets of Symbolic Regression for Scientific Discovery

NeurIPS 2022 AI for Science Workshop

Yoshitomo Matsubara, Naoya Chiba, Ryo Igarashi, Yoshitaka Ushiku

This work was mainly done while I was a research intern at OMRON SINIC X Corporation.

OpenReview Code SRSD-Feynman (Easy) SRSD-Feynman (Medium) SRSD-Feynman (Hard)