Spring 2023 RIT: Machine Learning For Rare Events

In recent years, machine learning techniques have penetrated a tremendous variety of scientific fields. In particular, they have given rise to data-driven methods for the study of rare event in complex physical systems such as conformational changes in biomolecules, rearrangements of clusters of interacting particles, etc. These methods truly opened new horizons by enabling us to address problems that used to be intractable due to the curse of dimensionality. In this RIT we will explore methods for the study of rare events based on machine learning.

Organizers: Dr. Maria Cameron, Margot Yuan, Shashank Sule


Optional. 1 credit. Course number: AMSC689, Section 0802. See Jessica Sadler for help with registration (jsadler@umd.edu). To earn 1 credit, you need to give a talk.


Time/Location: Fridays at 2:00 pm, Kirwan Hall 1310.

Feb 03: Maria Cameron, Model reduction and machine learning for quantifying rare events in stochastic systems (Organizational meeting). Slides

Feb 10: Margot Yuan, Causal disentanglement using causal graph regulation: autoencoders for learning image descriptions in simple terms

Feb 17: Zezheng Song, Finite expression method for solving PDEs. Application to the committor problem

March 10: JJ Guan, Physics-Informed Neural Networks for Numerical Solutions of PDEs.

March 31: Deep Ray, Bayesian inference using Generative Adversarial Networks.

April 7: Rachel Lee, A Bayesian Framework for Persistent Homology.

April 14: Frank McBride, Echo State Networks for Predicting Chaotic Time Series

May 5: Sam Potter, Recent progress on the butterfly factorization

Suggestions for papers:

You can give a talk on your research or on any paper of your choice if it is related to machine learning or rare events. Below are some suggested papers on model reduction and autoencoders:

Model reduction

  1. L. Maragliano, A. Fischer, E. Vanden-Eijnden, G. Ciccotti, String method in collective variables: minimum energy paths and isocommittor surfaces, 2006

  2. F. Legal and T. Lelievre, Effective dynamics using conditional expectations, Nonlinearity 2010

  3. I. Gyoengy, Mimicking the One-dimensional marginal distributions of processes having an Ito differential, 1986

  4. W. Zhang, C. Hartmann, C. Schuette, Effective dynamics along given reaction coordinate and reaction rate theory, 2016

  5. C. Hartmann, C. Schuette, W. Zhang, Model reduction algorithms for optimal control and importance sampling, 2016

  6. F. Nueske, P. Koltai, L. Boninsegna, C. Clementi, Spectral properties of effective dynamics from conditional expectations, 2021


  1. W. Chen, A. Ferguson, Molecular enhances sampling with auto encoders: on-the-fly collective variable discovery and accelerated free energy landscape exploration, 2018

  2. W. Chen, A. Tan, A. Ferguson, Collective variable discovery and enhanced sampling using auto encoders: innovations in network architecture and error function design, 2018

  3. C. Wehmeyer and F. Noe, time-lagged auto encoders: deep learning of slow collective variables from molecular kinetics, 2018

  4. Z. Belkacemi, P. Gkeka, T. Lelievre, G. Stoltz, Chasing collective variables using auto encoders and biased trajectories, 2022

  5. E. Crabtree, J. Bello-Rivas, A. Ferguson, I. Kevrekidis, GANs and closures: micro-macro consistency in multiscale modeling, 2022