Data science problems in statistical physics, computational statistics and machine learning share many similarities. However, the research is frequently carried out in isolation and analogous approaches and algorithms are often reinvented in each field separately.
The difficulty in communication and cross-fertilisation between the disciplines lies not only in their specific scientific jargon, but also in their research cultures and in the application-driven specification of problems that focuses on different desirable aspects of the solution, such as methodological rigour, convergence rates, efficiency, or scalability in different regimes.
This interest group will facilitate and stimulate the exchange of ideas, expertise, and perspectives on data science problems among these three fields.
John Aston, University of Cambridge
Gareth Roberts, University of Warwick
Andrew Duncan, University of Sussex