Artificial Intelligence & Machine Learning for Fundamental Science

Scientific motivation

In the past ten years or so the use of Artificial Intelligence and in particular Machine Learning has seen a dramatic increase in physics and astronomy. Image classification, signal processing and real-time data selection rely on the recognition of patterns in very large data sets, e.g., in the context of particle physics at the Large Hadron Collider or gravitational wave detection from neutron star/black hole mergers as seen by LIGO/Virgo. More recently, ML methods are also being developed to speed up numerical Monte Carlo approaches relying on high-performance computing (HPC), e.g., in numerical statistical mechanics and lattice Quantum Chromodynamics. Reversely, the methodology of statistical mechanics and the Monte Carlo approach (exactness, detailed balance, histogram reweighting, renormalization group) is being transferred to the ML community, to provide a fresh impetus to clarify the successes of the approach. Due to the similarities of questions asked, a cross-disciplinary approach in, e.g., image classification, links the physical and medical sciences, leading to unexpected connections between those usually separated disciplines. And finally, a longer-term goal is to fuse ML and quantum computing as quantum machine learning, which would lead to a dramatic speed-up and new, not yet explored, possibilities.

In order to build capacity, bridge gaps across disciplines, and develop a common framework, in this project we bring together researchers based in Korea and in the UK via three meetings: a virtual one, one hosted in Korea, and one hosted in the UK.

Practicalities for the first virtual meeting

• The meeting will run over three days: Monday 15/11, Tuesday 16/11 and Wednesday 17/11 2021.
• Every day the meeting will take place during 9am-11am (UK time) / 6-8pm (Korea time), using zoom.
• The purpose of the meeting is exchange scientific ideas and interests.
• To enable networking and discussion, we plan a short sets of talks for each day: 30+5' opening talk, three or four 15+5' talks, with further discussion in zoom breakout rooms.
• One aim of the meeting is to determine the scope of the two follow-up face-to-face meetings.
• This project is aligned to the activities taking place within AIMLAC, the UKRI Centre for Doctoral Training in Artificial Intelligence, Machine Learning & Advanced Computing.
• We gratefully thank our sponsors: the National Research Foundation of Korea, the British Embassy Seoul, and the UK Science & Innovation Network for this opportunity.
• For more information, please send an email to Seyong ( skim @ sejong.ac.kr) or Gert (g.aarts @ swanse.ac.uk).

Scientific programme

The programme can be found here.

List of participants