RIT: Machine Learning for rare events

Organizers: Maria Cameron and Luke Evans

Meetings: Friday, 12 PM

Room: MTH1310

Synopsis

In recent years, machine learning techniques penetrated a tremendous variety of scientific fields. In particular, they gave 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. They are divided into two families: diffusion map-based and neural network-based. In this RIT we will explore methods for the study of rare events based on machine learning.

Program

Each meeting, one of the participants will give a talk on a paper relevant to the subject of the RIT or on his/her research if it is related to ML or rare events.

Registration

There is an option for students to register for this RIT for one credit: AMSC689 section 0802. In order to register, students need to contact Jessica Sadler (jsadler at umd dot edu), the AMSC program coordinator, and provide their UID. Students who has registered are expected to give a talk.

Some Suggestions For Papers to Present

Schedule for Spring 2022

Organizational Meeting Slides

Maria Cameron
Friday February 4, 12PM

Modeling the fracture of hydrogels

Manyuan Tao
Friday February 11th, 12PM

Solving the committor equation using tensor trains

Margot Yuan
Friday February 25th, 12PM

Computing committors using Mahalanobis diffusion maps

Luke Evans
Friday March 4th, 12PM

AlphaZero

Max Springer
Friday March 18th, 12PM

Deep learning for solving backward stochastic differential equations

Zhirui Li
Friday April 1st, 12PM

Simulation for rare events in quantum error correction

Raley Roberts
Friday April 8th, 12PM

A new constructive heuristic driven by machine learning for the traveling salesman problem/Error analysis for target-measure diffusion maps

Shuhan Kou/Shashank Sule
Friday April 22nd, 12PM

Physics-Informed Neural Networks/The Deep Ritz Method

Lang Song/Jiaxing Liang
Friday April 29th, 12PM

Wasserstein gradient flows

Jiaqi Wang
Friday May 6th, 12PM

Schedule for Fall 2021

An overview. Slides

Maria Cameron
Friday September 10, 2PM

Diffusion maps applied to molecular dynamics

Luke Evans
Friday September 24, 2PM

Quantifying rare events with the aid of neural networks

Margot Yuan
Friday October 8, 2pm

Analysis of activation functions for neural networks (based on A. Townsend's papers)

Manyuan Tao
Friday October 15, 2pm

Self-assembly of hydrocarbons

Christopher Moakler
Friday October 22, 2pm

Mean-field analysis and scaling limits for neural networks (based on works by J. Sirignano and K. Spiliopoulos) (cancelled)

Maria Cameron

Friday November 5, 2pm

An Introduction to Density Functional Theory. Slides

Ryan Synk
Friday November 19, 2pm

Spectralnet

Shashank Sule
Friday December 3, 2pm