Bridging machine learning and statistical field theory through generative and diffusion processes

Recent years have seen an explosion of cross-disciplinary research linking machine learning and theoretical physics. In this workshop, we focus on generative methods as applied to lattice field theory, with QCD as a long-term target. With many approaches being explored at various levels of maturity (normalising flow, including stochastic and continuous normalising flow, diffusion models, transport models, ...) and a deepening understanding from the perspective of theoretical physics (links to the renormalisation group, gauge symmetries, stochastic quantisation, random matrix theory, ...), it is time to take stock and constructively assess the applicability of the methodology and future perspectives in the field. The meeting will bring together researchers from the theoretical physics community, active in lattice field theory and statistical physics, with researchers in ML, to optimise the possibility of knowledge exchange. As the field is evolving rapidly, we take care to invite proponents of new and promising methods, as they are developed between the submission of the proposal and the delivery of the workshop. The key outcome will be the advancement of a common foundation and the agreement on a set of targets for the short and intermediate term.