Indo-French Workshop on Navigating rugged landscapes

Title: Indo-French Workshop on Navigating rugged landscapes (NRL): Machine learning approach to dynamics and enhanced sampling in disordered systems, and statistical physics approaches to machine learning

About the Workshop:

The workshop aims to bring together experts working in seemingly different domains of glassy dynamics, biophysics, and machine learning to address the common underlying threads; investigating the rugged landscape to learn the behavior of dynamical systems. This will lead to the exchange ideas across domains initiating discussions on disruptive ideas, for instance, (i) similarities between the loss landscapes of machine learning and glassy landscapes, (ii) relationships between the stability of neural networks to adversarial attacks and the perturbation in glassy landscapes, (iii) can generative models for out-of-distribution samples be used for enhanced sampling of rare events, and (iv) can the glass structure and dynamics be characterised by learnable and interpretable low-dimensional features. The workshop will systematically address these questions and initiate discussions to narrow down and identify the key problems to be solved.

The workshop has three major themes under which the discussions will be organised. These, identified

Theme I:  Machine learning investigation of glasses and other disordered systems

Theme II: Enhanced Sampling with Machine Learning

Theme III: Statistical Physics of Machine Learning

Workshop Webpage: https://sites.google.com/view/nrl2025