Organizing Committee
Abstract

Note: At this time, ICERM is no longer accepting applications for this workshop as we are at capacity. Talks will be live streamed and recorded for viewing.

The machine learning revolution is already having a significant impact across the social sciences and business, but it is also beginning to change computational science and engineering in fundamental and very varied ways.

We are experiencing the rise of new and simpler data-driven methods based on techniques from machine learning such as deep learning. This revolution allows for the development of radical new techniques to address problems known to be very challenging with traditional methods and suggests the potential dramatic enhancement of existing methods through data informed parameter selection, both in static and dynamic modes of operation. Techniques are emerging that allows us to produce realistic solutions from non-sterilized computational problems in diverse physical sciences.

However, the urgent and unmet need to formally analyze, design, develop and deploy these emerging methods and develop algorithms must be addressed. Many central problems, e.g., enforcement of physical constraints in machine learning techniques and efficient techniques to deal with multiscale problems, are unmet in existing methods.

The primary goal of this Hot Topic workshop is to bring together leading researchers across various fields to discuss recent results and techniques at the interface between traditional methods and emerging data-driven techniques to enable innovation in scientific computing in computational science and engineering.

This workshop is fully funded by a Simons Foundation Targeted Grant to Institutes.

Image for "Scientific Machine Learning"

Confirmed Speakers & Participants

  • Speaker
  • Poster Presenter
  • Attendee
  • Virtual Attendee

Workshop Schedule

Monday, January 28, 2019
TimeEventLocationMaterials
8:00 - 8:55am ESTRegistration - ICERM 121 South Main Street, Providence RI 0290311th Floor Collaborative Space 
8:55 - 9:00am ESTWelcome - ICERM Director11th Floor Lecture Hall 
9:00 - 9:45am ESTMachine learning and multi-scale modeling - Weinan E, Princeton University11th Floor Lecture Hall
9:55 - 10:40am ESTWhich ReLU Net Architectures Give Rise to Exploding and Vanishing Gradients - Boris Hanin, Texas A&M & Facebook11th Floor Lecture Hall
10:40 - 11:00am ESTCoffee/Tea Break11th Floor Collaborative Space 
11:00 - 11:45am ESTDeep Neural Network, Finite Element and Multigrid - Jinchao Xu, The Penn State University11th Floor Lecture Hall
11:55 - 12:40pm ESTA new perspective on machine learning - Hrushikesh Mhaskar, Claremont Graduate University (Claremont, CA, US)11th Floor Lecture Hall
12:40 - 2:00pm ESTBreak for Lunch / Free Time  
2:00 - 2:45pm ESTHidden Physics Models - Machine Learning of Non-Linear Partial Differential Equations - Maziar Raissi, Brown University11th Floor Lecture Hall
2:55 - 3:40pm ESTCollapse of deep and narrow neural nets - Lu Lu, Brown University & Yeonjong Shin, Brown University11th Floor Lecture Hall
3:40 - 4:00pm ESTCoffee/Tea Break11th Floor Collaborative Space 
4:00 - 5:00pm ESTDiscussion 11th Floor Lecture Hall 
5:00 - 6:30pm ESTWelcome Reception11th Floor Collaborative Space 
Tuesday, January 29, 2019
TimeEventLocationMaterials
9:00 - 9:45am ESTDeep Neural Networks and Partial Differential Equations - Approximation Theory and Structural Properties - Philipp Petersen, University of Oxford11th Floor Lecture Hall
9:55 - 10:40am ESTData-driven model discovery and coordinate embeddings for physical systems - Nathan Kutz, University of Washington11th Floor Lecture Hall
10:40 - 11:00am ESTCoffee/Tea Break11th Floor Collaborative Space 
11:00 - 11:45am ESTNonlinear reduced-order modeling - Using machine learning to enable extreme-scale simulation for many-query problems - Kevin Carlberg, Sandia National Laboratories11th Floor Lecture Hall
11:55 - 12:40pm ESTNo equations, no variables, no parameters, no space and no time - some new and some old results in data-driven modeling of complex dynamical systems - Yannis Kevrekidis, Johns Hopkins University11th Floor Lecture Hall
12:40 - 12:50pm ESTGroup Photo11th Floor Lecture Hall 
12:50 - 2:15pm ESTBreak for Lunch / Free Time  
2:15 - 3:00pm ESTData Driven Governing Equations Approximation Using Deep Neural Networks - Dongbin Xiu, Ohio State University11th Floor Lecture Hall
3:15 - 4:00pm ESTEnforcing constraints for interpolation and extrapolation in Generative Adversarial Networks - Panos Stinis, Pacific Northwest National Laboratory11th Floor Lecture Hall
4:00 - 6:00pm ESTPoster Session - Includes: Coffee Break11th Floor Collaborative Space 
Wednesday, January 30, 2019
TimeEventLocationMaterials
9:00 - 9:45am ESTDeep Neural Networks for Data-Driven Turbulence Models - Andrea Beck, University of Stuttgart11th Floor Lecture Hall
9:55 - 10:40am ESTSequential Particle Flow for Bayesian Inference - Le Song, Georgia Institute of Technology11th Floor Lecture Hall
10:40 - 11:00am ESTCoffee/Tea Break11th Floor Collaborative Space 
11:00 - 11:45am ESTSolid Harmonic Wavelet Scattering for Predicting Molecule Properties - Michael Eickenberg, UC Berkeley11th Floor Lecture Hall
11:55 - 12:40pm ESTEstimating Learnability - Gregory Valiant, Stanford University11th Floor Lecture Hall
12:40 - 2:00pm ESTBreak for Lunch / Free Time  
2:00 - 2:45pm ESTProject Artie - An Artificial Student for Disciplines Informed by Partial Differential Equations - Anthony Patera, MIT11th Floor Lecture Hall
2:55 - 3:10pm ESTClosing Remarks11th Floor Lecture Hall 

Lecture Videos

Estimating Learnability

Gregory Valiant
Stanford University
January 30, 2019

Deep Neural Networks for Data-Driven Turbulence Models

Andrea Beck
University of Stuttgart
January 30, 2019

Collapse of deep and narrow neural nets

Lu Lu
Brown University
Yeonjong Shin
Brown University
January 28, 2019

A new perspective on machine learning

Hrushikesh Mhaskar
Claremont Graduate University
January 28, 2019

Deep Neural Network, Finite Element and Multigrid

Jinchao Xu
Pennsylvania State University
January 28, 2019

Machine learning and multi-scale modeling

Weinan E
Princeton University
January 28, 2019