Hot Topic: The Machine-Learning Revolution – Part 1: Mathematics

Sanjeev Arora, Maia Fraser, Javier Gómez-Serrano, Yang-Hui He

It’s hard to keep up with all the tools that machine learning offers mathematics and theoretical science. It’s also hard to avoid the hype that tends to swirl around ML. Two panels of mathematicians and scientists who have pioneered the use of ML in their disciplines will discuss how they navigate these challenges. 

The first panel will focus on mathematics research, in which proof assistants are enabling collaborations of a scale never before seen, while experimental mathematics and pattern-discovery algorithms are revealing new links between seemingly unrelated areas. 

The second panel will focus on physical science, where ML provides fresh methods of simulation, of data analysis, and of effective descriptions of phenomena.

But some worry that these advances are not unequivocally good. ML may exacerbate biases in the data, shift how the research community assigns value to problems, give tech companies undue influence over pure research, and steer science away from its traditional goal of low-dimensional explanation. Will humans continue to be the protagonists of discovery, or will we one day be reduced to training and querying models like modern-day oracles? 

Moderator:
George Musser 

Panelists:
Sanjeev Arora (ACM Prize in Computing 2011)
Maia Fraser
Javier Gómez-Serrano
Yang-Hui He

---