Challenges and Prospects of ML for the Physical Sciences

America/New_York
Ingrid Daubechies Auditorium/2-IDA (162 5th Avenue)

Ingrid Daubechies Auditorium/2-IDA

162 5th Avenue

Flatiron Institute Entrance on 21st St.
200
Description
Our goal is to gather a diverse community of researchers at the cross-roads of ML, Statistics, Applied Mathematics and Physics to exchange views and ideas on current successes and roadblocks of data-driven tools applied to computational science. 
Wi-Fi Network: SimonsGuests Password: simonsnyc
    • 8:30 AM 9:00 AM
      Breakfast 30m 162 Fifth Avenue/2-2nd Floor Promenade

      162 Fifth Avenue/2-2nd Floor Promenade

      162 5th Avenue

      200
    • 9:00 AM 9:30 AM
      Joan Bruna 30m

      Exponential Separations in Symmetric and Anti-Symmetric Neural Networks

      Speaker: Joan Bruna (NYU / CCM)
    • 9:30 AM 9:45 AM
      Q&A
    • 9:45 AM 10:15 AM
      Alberto Bietti 30m

      Benefits of convolutional models: a kernel perspective

      Speaker: Alberto Bietti (NYU)
    • 10:15 AM 10:30 AM
      Q&A
    • 10:30 AM 11:00 AM
      Coffee Break 30m
    • 11:00 AM 11:30 AM
      Yi Ma 30m

      CTRL: Closed-Loop Data Transcription via Rate Reduction

      Speaker: Yi Ma (University of California, Berkeley)
    • 11:30 AM 11:45 AM
      Q&A
    • 11:45 AM 12:15 PM
      Soledad Villar 30m

      Dimensionless machine learning: Imposing exact units equivariance

      Speaker: Soledad Villar (Johns Hopkins University)
    • 12:15 PM 12:30 PM
      Q&A
    • 12:30 PM 2:00 PM
      Lunch 1h 30m 162 Fifth Avenue/2-2nd Floor Promenade

      162 Fifth Avenue/2-2nd Floor Promenade

      162 5th Avenue

      200
    • 2:00 PM 2:30 PM
      CCN - Eero Simoncelli 30m

      Image Priors in the Era of Machine Learning

    • 2:30 PM 3:00 PM
      Jiequn Han 30m

      Developing reduced-order PDEs with machine learning-based closure models

      Speaker: Jiequn Han (CCM, Flatiron Institute)
    • 3:00 PM 3:15 PM
      Q&A
    • 3:15 PM 3:45 PM
      Coffee Break 30m
    • 3:45 PM 5:00 PM
      Working Session 1h 15m
    • 8:30 AM 9:00 AM
      Breakfast 30m 162 Fifth Avenue/2-2nd Floor Promenade

      162 Fifth Avenue/2-2nd Floor Promenade

      162 5th Avenue

      200
    • 9:00 AM 9:30 AM
      Lawrence Saul 30m

      The method of sparse similarity matching for high dimensional data analysis

      Speaker: Lawrence Saul (UCSD/CCM )
    • 9:30 AM 9:45 AM
      Q&A
    • 9:45 AM 10:15 AM
      SueYeon Chung 30m

      Efficient representation geometry in distributed neural networks

      Speaker: SueYeon Chung (CCN)
    • 10:15 AM 10:30 AM
      Q&A
    • 10:30 AM 11:00 AM
      Coffee Break 30m
    • 11:00 AM 11:30 AM
      Rose Yu 30m

      Approximately Equivariant Networks for Imperfectly Symmetric Dynamics

      Speaker: Rose Yu (UCSD)
    • 11:30 AM 11:45 AM
      Q&A
    • 11:45 AM 12:15 PM
      Léon Bottou 30m

      Out-of-distribution generalization and causation

      Speaker: Léon Bottou (Facebook)
    • 12:15 PM 12:30 PM
      Q&A
    • 12:30 PM 2:00 PM
      Lunch 1h 30m
    • 2:00 PM 2:30 PM
      CCB - Mike Shelley 30m
    • 2:30 PM 3:00 PM
      Michael Mahoney 30m

      Continuous Network Models for Sequential Predictions

      Speaker: Michael Mahoney (UC Berkeley )
    • 3:00 PM 3:15 PM
      Q&A
    • 3:15 PM 3:45 PM
      Kyle Cranmer 30m
      Speaker: Kyle Cranmer (NYU)
    • 3:45 PM 4:00 PM
      Q&A
    • 4:00 PM 4:30 PM
      Coffee Break 30m
    • 4:30 PM 5:00 PM
      Working Session 30m
    • 8:30 AM 9:00 AM
      Breakfast 30m 162 Fifth Avenue/2-2nd Floor Promenade

      162 Fifth Avenue/2-2nd Floor Promenade

      162 5th Avenue

      200
    • 9:00 AM 9:30 AM
      Maarten de Hoop 30m

      Deep learning, active tectonics and planetary exploration

      Speaker: Maarten de Hoop (Rice University )
    • 9:30 AM 9:45 AM
      Q&A
    • 9:45 AM 10:15 AM
      Marylou Gabrié 30m

      Enhancing Sampling with Learning

      Speaker: Marylou Gabrié (Ecole Polytechnique)
    • 10:15 AM 10:30 AM
      Q&A
    • 10:30 AM 11:00 AM
      Coffee Break 30m
    • 11:00 AM 11:30 AM
      Shirley Ho 30m

      Interpretable Machine Learning for Astrophysics

      Speaker: Shirley Ho (CCA )
    • 11:30 AM 11:45 AM
      Q&A
    • 11:45 AM 12:15 PM
      Giuseppe Carleo 30m

      Variational quantum states in the age of machine learning

      Speaker: Giuseppe Carleo (École Polytechnique)
    • 12:15 PM 12:30 PM
      Q&A
    • 12:30 PM 2:00 PM
      Lunch 1h 30m
    • 2:00 PM 2:30 PM
      CCA - David Hogg 30m
    • 2:30 PM 3:00 PM
      Wenda Zhou 30m

      Spatial equivariance and deep networks: practical approaches and challenges

      Speaker: Wenda Zhou (NYU/CCM )
    • 3:00 PM 3:15 PM
      Q&A
    • 3:15 PM 3:45 PM
      Tommaso Biancalani 30m

      Deep learning for virtual screens of antibiotics

      Speaker: Tommaso Biancalani (Genentech )
    • 3:45 PM 4:00 PM
      Q&A
    • 4:00 PM 4:30 PM
      Coffee Break 30m
    • 4:30 PM 5:30 PM
      Working Session 1h
    • 5:30 PM 8:00 PM
      Dinner at BLACKBARN Restaurant 2h 30m

      19 E 26th St, New York, NY 10010

    • 8:30 AM 9:00 AM
      Breakfast 30m 162 Fifth Avenue/2-2nd Floor Promenade

      162 Fifth Avenue/2-2nd Floor Promenade

      162 5th Avenue

      200
    • 9:00 AM 9:30 AM
      Robert Gower 30m

      Adaptive Stochastic Gradient Methods that leverage Interpolation

      Speaker: Robert Gower (CCM )
    • 9:30 AM 9:45 AM
      Q&A
    • 9:45 AM 10:15 AM
      Eric Vanden Eijnden 30m

      Probability flow solution of the Fokker-Planck equation

      Speaker: Eric Vanden Eijnden (NYU)
    • 10:15 AM 10:30 AM
      Q&A
    • 10:30 AM 11:00 AM
      Coffee Break 30m
    • 11:00 AM 11:30 AM
      Leslie Greengard 30m

      Electromagnetic/acoustic design optimization

      Speaker: Leslie Greengard (CCM )
    • 11:30 AM 11:45 AM
      Q&A
    • 11:45 AM 12:15 PM
      Yongji Wang 30m

      Physics-informed Neural Networks for fluid dynamics

      Speaker: Yongji Wang (Princeton University )
    • 12:15 PM 12:30 PM
      Q&A
    • 12:30 PM 2:00 PM
      Lunch 1h 30m
    • 2:00 PM 2:30 PM
      CCQ - Domenico Di Sante 30m
    • 2:30 PM 3:00 PM
      Miles Cranmer 30m

      One Trick to Improve Regularized Training of Neural Networks

      Speaker: Miles Cranmer (Princeton University/CCA)
    • 3:00 PM 3:15 PM
      Q&A
    • 3:15 PM 3:45 PM
      Stéphane Mallat 30m

      Learning and Generating Multiscale Physics

      Speaker: Stéphane Mallat (Collège de France/CCM )
    • 3:45 PM 4:00 PM
      Q&A
    • 4:00 PM 5:00 PM
      Working Session 1h
    • 8:30 AM 9:00 AM
      Breakfast 30m
    • 9:00 AM 11:00 AM
      Working Session 2h
    • 11:00 AM 11:30 AM
      Lunch to-go 30m