Machine Learning Potentials - StAtus and FuturE (MLP-SAFE)

Institute for Computational and Data Sciences

Workshop for the community to learn from thought leaders  about the state of the art and future of machine learning potentials

T﻿hank you all for a wonderful workshop! Presenters slides can be found <a href="https://drive.google.com/drive/folders/1nLjN4TJ0mZ9U7S9hljoTbHraOWYczmNA?usp=sharing" target="_blank" data-msys-clicktrack="0" rel="nofollow noopener noreferrer">here</a>. SPECIAL NOTE - we have had an overwhelming positive response to this workshop and have unfortunately hit our web service limits. Therefore, we are capping registration at 750. We will be able to host up to 500 attendees at one time. If you attempt to join and cannot get in please try again a bit later or for the next session. If you are unable to register - please join the waitlist and we can let you know if there are spaces as each session starts.The Machine Learning Potentials - StAtus and FuturE (MLP-SAFE) workshop is taking place virtually from July 17-19 from 11am-3pm US EDT . MLPs are rapidly transforming materials simulation and discovery. This workshop will provide an outstanding opportunity to efficiently integrate knowledge and share with the community about key advancements and opportunities in MLPs. The workshop will include six 2 hour sessions, each with typically four 15 min presentations from thought leaders, followed by an hour of panel style Q&amp;A. The session topics and approximate schedule are given below (additional d﻿etails will be added as speakers are confirmed).The key ideas from the meetings will be integrated in a report publication to help guide researchers and funding agencies.Organized by Dane Morgan (Dept of Materials Science and Engineering, University of Wisconsin-Madison), Boris Kozinsky (School of Engineering and Applied Sciences, Harvard University), and Vasant Honavar (College of Information Sciences and Technology, Penn State University). Supported by the National Science Foundation through the Institute for AI-Enabled Materials, Discovery, Design, and Synthesis (AIMS) planning grant # 2020243.<h2></h2><h2>AGENDA</h2>A﻿ few general items<ol><li>Except for the plenary talk and first session, each session will be approximately one hour of talks followed by one hour of discussion. ~10min will be taken at the end of first session each day for a break. </li><li>All materials will be recorded for internal use only.</li></ol><h3>Day 1 - Monday 17 July</h3>11:00am to 11:15am - Welcome &amp; Introductions11:15am to 12:15pm - Plenary Overview Presentation<ul><li>Gabor Csanyi, Professor, University of Cambridge - Machine learning brought a revolution in force field</li></ul>12:15pm to 1:00pm - Session 1What is vision for AI/ML in materials/chemistry in government programs? Confirmed speakers:<ul><li>Cosima Boswell-Koller, Program Director, Mathematical and Physical Science (MPS), National Science Foundation - The Role of AI/ML in NSF’s Division of Materials Research</li><li>Sylvia Spengler, Program Director, Computer and Information Science and Engineering (CISE), National Science Foundation - AI/ML and Materials from the NSF/CISE Perspective</li><li>Marco Fornari, Program Manager, Advanced Scientific Computing Research · U.S. Department of Energy (DOE)</li></ul>1:00pm to 3:00pm - Session 2What are types of MLPs and what are pros/cons of different types? What have we learned about what works? Try to discuss in terms of feature space, model, and training approaches.Confirmed speakers:<ul><li>Nongnuch Artrith, Professor, Universiteit Utrecht</li><li>Ralf Drautz, Professor, Ruhr-Universität Bochum - Expanding atomic interaction models</li><li>Olexandr Isayev, Professor, Carnegie Mellon University</li><li>Boris Kozinsky, Professor, Harvard University</li></ul><h3>Day 2 - Tuesday 18 July</h3>11:00am to 1:00pm - Session 3What are types of MLPs and what are pros/cons of different types? What have we learned about what works? Try to discuss in terms of feature space, model, and training approaches.Confirmed speakers:<ul><li>Shyue Ping Ong, Professor, UC San Diego - Code, Data and Infrastructure Gaps in Machine Learning Potentials for Materials Science</li><li>So Takamoto, Researcher, Preferred Networks, Inc. - Tensorial message passing based graph neural network for universal interatomic potential</li><li>Lingfeng Zhang, Chief scientist of DP Technology, Researcher at AI for Science Institute, Beijing - Machine Learning Potentials at Scale</li><li>Aidan Thompson, Distinguished Member of Technical Staff, Sandia National Laboratories - ML Potentials in LAMMPS</li></ul>1:00pm to 3:00pm - Session 4What physics is missing (e.g., magnetism, long-range forces)?Confirmed speakers:<ul><li>Cecilia Clementi, Professor, Freie Universität Berlin</li><li>Bowen Deng, University of California, Berkeley - CHGNet: pretrained universal neural network potential for charge-informed atomistic modeling</li><li>I﻿gor Polavsky, Research Scientist, University of Luxembourg - Fully Quantum (Bio)Molecular Simulations: Dream or Reality?</li><li>Julia Westermayr, Professor, Universität Leipzig - Machine learning electronically excited states</li></ul><h3>Day 3 - Wednesday 19 July</h3>11:00am to 1:00pm - Session 5What infrastructure is needed (e.g., test datasets, potential codes)?Confirmed speakers:<ul><li>K﻿amal Choudhary, Scientist, Thermodynamics and Kinetics Group - National Institute of Standards and Technology (NIST) - JARVIS-Leaderboard Benchmarks &amp; Unified Graph Neural Network Force-Field</li><li>Anders Johansson, Harvard University - Autograd vs. elbow grease: Comparing Allegro and FLARE for performance-portable extreme-scale simulations</li><li>Stefano Martiniani, Professor, New York University</li><li>K﻿J Schmidt, Machine Learning Research Scientist at the University of Chicago</li><li>Brandon Wood, Research Scientist, Meta AI - The 2023 Open Catalyst Challenge </li></ul>1:00pm to 3:00pm - Session 6Where do we need to go next as a community (dream big) (e.g., very large datasets, universal force fields, new ML approaches?)<ul><li>E﻿kin Dogus Cubuk, Research Scientist, Google DeepMind - Scaling up materials discovery using machine learning: how to evaluate useful generalization?</li><li>X﻿iang Fu, Massachusetts Institute of Technology - The Journey of ML Force Fields: From Accuracy to Stability, Efficiency, and Extended Scales</li><li>Jonathan Godwin, Co-founder, Orbital Materials (ex-Google Deepmind ) - Are symmetries all we need? Directions beyond equivariance in GNNs</li><li>A﻿nuj Karpatne, Professor, Virginia Tech - Knowledge-guided Machine Learning: Advances in An Emerging Paradigm Combining Scientific Knowledge with Machine Learning</li></ul>

Our {communityGuidelinesLink} describe the sort of content we prohibit on Eventbrite. If you suspect an event may be in violation, you can report it to us so we can investigate.

To report other categories of prohibited or illegal content, submit {link}

Machine Learning Potentials - StAtus and FuturE (MLP-SAFE)

More events from Institute for Computational and Data Sciences

Discover more events from Institute for Computational and Data Sciences, from Science & Tech to other experiences you might love.

Still looking for the right event?

Explore all online events to browse and filter by date, category, and more.

Machine Learning Potentials - StAtus and FuturE (MLP-SAFE)

More events from Institute for Computational and Data Sciences

Discover more events from Institute for Computational and Data Sciences, from Science & Tech to other experiences you might love.

You might also like...

Browse more events with different dates, prices, and formats to find your next great experience.

Still looking for the right event?

Explore all online events to browse and filter by date, category, and more.