Workshop IV: Using Physical Insights for Machine Learning - IPAM

Workshop IV: Using Physical Insights for Machine Learning

Part of the Long Program Machine Learning for Physics and the Physics of Learning
November 18 - 22, 2019

Schedule

All times in this Schedule are Pacific Time (PT)

Monday, November 18, 2019

Morning Session

08:00-08:55
Check-In/Light Breakfast (Hosted by IPAM)
08:55-09:00
Welcome & Opening Remarks: Dean Miguel García-Garibay (Dean of Physical Sciences, UCLA) and Dima Shlyakhtenko (Director, IPAM)
09:00-09:50
Lenka Zdeborová (Commissariat à l'Énergie Atomique (CEA))
Understanding machine learning via exactly solvable statistical physics models
10:00-10:15
Break
10:15-11:05
Rémi Monasson (Centre National de la Recherche Scientifique (CNRS))
Capacity-resolution trade-off in the optimal learning of multiple low-dimensional manifolds by attractor neural networks
11:15-11:30
Break
11:30-12:20
Eric Vanden-Eijnden (Courant Institute of Mathematical Sciences)
Trainability and accuracy of artificial neural networks

Afternoon Session

12:30-14:30
Lunch (on your own)
14:30-15:20
Risi Kondor (University of Chicago & Flatiron Institute)
Covariant neural network architectures for learning physics
15:30-15:45
Break
15:45-16:35
Michele Ceriotti (MLPRC2)
Machine learning for atomic and molecular simulations
16:45-17:15
Lightning Poster Presentations
17:15-18:30
Poster Session & Reception (Hosted by IPAM)

Tuesday, November 19, 2019

Morning Session

08:00-09:00
Check-In/Breakfast (Hosted by IPAM)
09:00-09:50
Yann LeCun (New York University)
Energy-Based Self-Supervised Learning
10:00-10:15
Break
10:15-11:05
Gérard Ben Arous (New York University)
Which is worse: a weak signal lost in entropy or trapping by topological complexity?
11:15-11:30
Break
11:30-12:20
Chiara Cammarota (King's College London)
High-dimensional cost landscape and gradient descent in Tensor PCA and its generalisations

Afternoon Session

12:30-14:30
Lunch (on your own)
14:30-15:20
Masayuki Ohzeki (Tohoku University)
Quantum annealing and machine learning - new directions of quantum
15:30-16:00
Break
16:00-16:50
Haim Sompolinsky (The Hebrew University of Jerusalem)
Statistical Mechanics of Deep Manifolds: Mean Field Geometry in High Dimension

Wednesday, November 20, 2019

Morning Session

08:00-09:00
Check-In/Breakfast (Hosted by IPAM)
09:00-09:50
Riccardo Zecchina (Bocconi University)
Evidence of local entropy optimization in machine learning, physics and neuroscience
10:00-10:15
Break
10:15-11:05
Daniel Roberts (Diffeo)
Deep learning as a toy model of the 1/N-expansion and renormalization
11:15-11:30
Break
11:30-12:20
Carlo Baldassi (Bocconi University)
On the existence of wide flat minima in neural network landscapes: analytic and algorithm approaches

Afternoon Session

12:30-14:30
Lunch (on your own)
14:30-15:20
Maria Schuld (University of KwaZulu-Natal)
Innovating machine learning with near-term quantum computing
15:30-16:00
Break
16:00-16:50
Maxwell Hutchinson (Citrine Informatics)
Boltzmann Trees: a physically inspired randomization for robust modeling of physical data

Thursday, November 21, 2019

Morning Session

08:00-09:00
Check-In/Breakfast (Hosted by IPAM)
09:00-09:50
Andrea Montanari (Stanford University)
Thermodynamic limits for neural networks
10:00-10:15
Break
10:15-11:05
Marc Mezard (Ecole Normale Supérieure)
Machine learning with neural networks: the importance of data structure
11:15-11:30
Break
11:30-12:20
Soledad Villar (New York University)
Graph neural networks for combinatorial optimization problems on graphs

Afternoon Session

12:30-14:30
Lunch (on your own)
14:30-15:30
Chris Sutton (Fritz-Haber-Institut der Max-Planck-Gesellschaft)
Subgroup Discovery for Assessing the Domain of Applicability of Machine Learning Models
15:30-16:00
Break
16:00-16:50
Matthias Rupp (MLPRC2)
How to assess scientific machine learning models? Prediction errors and predictive uncertainty quantification
Printable Version