For the first time, AI helps mathematicians

2 December 2021

Mathematicians have obtained new insights in this sector as a result of machine learning methods, and one of the three major authors of the research is Dr. Petar Veličković, a member of the Institute for Artificial Intelligence's Advisory Board.

The research was conducted in partnership with the University of Oxford, the University of Sydney in Australia, and DeepMind, Google's artificial intelligence sister business.

DeepMind collaborated with mathematicians from the universities of Oxford and Sydney to investigate the potential of machine learning to discern mathematical structures and patterns. For the first time, the application of AI systems was critical to the development of a new theorem in the domains of knot theory and representation theory.

For the first time, AI helps mathematicians

Author: Nature

The findings are detailed in a new report published in Nature, the world's leading scientific journal. The paper is featured on the cover of the most recent edition of Nature.

This study was completed by a team that included DeepMind's Dr. Petar Veliković and Dr. Nenad Tomašev. Dr Veliković commented on the study on Twitter, writing, “(G)NNs can successfully guide the intuition of mathematicians & yield top-tier results -- in both representation theory & knot theory.”

The authors believe that their study can serve as a paradigm for strengthening collaboration across the domains of mathematics and artificial intelligence in order to generate unanticipated discoveries by exploiting the respective strengths of mathematics and machine learning.

Read the published paper at the following link, and find out what the director of the University of Sydney's Mathematical Institute has to say about the report's results here.