A Perspective on Symbolic Machine Learning in Physical Sciences

TL;DR

This paper advocates for the integration of symbolic machine learning with numerical methods to enhance scientific discovery in physics, emphasizing interpretability and the dual nature of physical sciences.

Contribution

It highlights the importance of developing symbolic machine learning approaches alongside numerical methods for physics research, addressing interpretability issues.

Findings

Symbolic ML complements numerical ML in physics.

Interpretability is crucial for scientific discovery.

Parallel development of symbolic ML can accelerate physics research.

Abstract

Machine learning is rapidly making its pathway across all of the natural sciences, including physical sciences. The rate at which ML is impacting non-scientific disciplines is incomparable to that in the physical sciences. This is partly due to the uninterpretable nature of deep neural networks. Symbolic machine learning stands as an equal and complementary partner to numerical machine learning in speeding up scientific discovery in physics. This perspective discusses the main differences between the ML and scientific approaches. It stresses the need to develop and apply symbolic machine learning to physics problems equally, in parallel to numerical machine learning, because of the dual nature of physics research.