Physics of Machine Learning I
FOCUS · F09 · ID: 46543
Presentations
-
Toward Statistical Mechanics of Deep Learning
ORAL · Invited
–
Publication: Qianyi Li, and Haim Sompolinsky (2021). Statistical Mechanics of Deep Linear Neural Networks: The Back-Propagating Kernel Renormalization. (Physical Review X 11.3 031059) .
Presenters
-
Haim I Sompolinsky
- The Hebrew University of Jerusalem and Harvard University
- Hebrew University of Jerusalem
- Center for Brain Science, Harvard Univer
Authors
-
Haim I Sompolinsky
- The Hebrew University of Jerusalem and Harvard University
- Hebrew University of Jerusalem
- Center for Brain Science, Harvard Univer
-
Qianyi Li
- Biophysics Program, Harvard University
-
-
AI Pontryagin or: How Artificial Neural Networks Learn to Control Dynamical Systems
ORAL
–
Publication: This work is currently under review in Nature Communications. A second, related work is under review in Physical Review Research.
Presenters
-
Lucas Boettcher
- Frankfurt School of Finance and Management; UCLA
- Frankfurt School of Finance & Management gGmbH
Authors
-
Lucas Boettcher
- Frankfurt School of Finance and Management; UCLA
- Frankfurt School of Finance & Management gGmbH
-
Thomas Asikis
- ETH Zurich
-
Nino Antulov-Fantulin
- ETH Zurich
-
-
Probing the Theoretical and Computational Limits of Dissipative Design
ORAL
–
Publication: Shriram Chennakesavalu and Grant M. Rotskoff. Probing the theoretical and computational limits of dissipative design, 2021. arXiv: 2108.05452 [cond-mat.stat-mech].
Presenters
-
Shriram Chennakesavalu
- Stanford University
Authors
-
Shriram Chennakesavalu
- Stanford University
-
Grant M Rotskoff
- Stanford Univ
-
-
The Role of Data in the Sloppiness of Deep Networks
ORAL
–
Presenters
-
Pratik Chaudhari
- University of Pennsylvania
Authors
-
Pratik Chaudhari
- University of Pennsylvania
-
Rubing Yang
- University of Pennsylvania
-
Jialin Mao
- University of Pennsylvania
-
-
Statistical Mechanics of Kernel Regression and Wide Neural Networks
ORAL
–
Publication: Blake Bordelon, Abdulkadir Canatar, Cengiz Pehlevan, Spectrum Dependent Learning Curves in Kernel Regression and Wide Neural Networks, ICML, 2020
Abdulkadir Canatar, Blake Bordelon, Cengiz Pehlevan, Spectral bias and task-model alignment explain generalization in kernel regression and infinitely wide neural networks, Nature Communications, 2021
Abdulkadir Canatar, Blake Bordelon, Cengiz Pehlevan, Out-of-Distribution Generalization in Kernel Regression, NeurIPS, 2021Presenters
-
Abdulkadir Canatar
- Harvard University
Authors
-
Abdulkadir Canatar
- Harvard University
-
Blake Bordelon
- Harvard University
-
Cengiz Pehlevan
- Harvard University
-
-
Machine learning in and out of equilibrium
ORAL
–
Presenters
-
Michael Hinczewski
- Case Western Reserve University
Authors
-
Michael Hinczewski
- Case Western Reserve University
-
Shishir Adhikari
- Harvard Medical School
-
Alkan Kabakcioglu
- Koc University
- Koç University
-
Alexander Strang
- University of Chicago
-
Deniz Yuret
- Koç University
-
-
Can Artificial Intelligence "formulate" Quantum Mechanics? An Illustration for Planck's Blackbody Radiation
ORAL
–
Publication: In preparation:
V. Shankar, S. Shankar, "Can Artificial Intelligence "formulate" Quantum Mechanics? An Illustration for Planck's Blackbody Radiation"Presenters
-
Vishnu Shankar
- Stanford University
Authors
-
Vishnu Shankar
- Stanford University
-
Sadasivan Shankar
- SLAC National Laboratory and Stanford University
- Harvard University
-
-
Non-Gaussian effects in finite Bayesian neural networks
ORAL
–
Publication: Preprints: https://arxiv.org/abs/2104.11734, https://arxiv.org/abs/2106.00651
Presenters
-
Jacob Zavatone-Veth
- Harvard University
Authors
-
Jacob Zavatone-Veth
- Harvard University
-
Abdulkadir Canatar
- Harvard University
-
Benjamin S Ruben
- Harvard University
-
Cengiz Pehlevan
- Harvard University
-
-
Equilibrium and non-Equilibrium regimes in the learning of Restricted Boltzmann Machines
ORAL
–
Publication: https://arxiv.org/pdf/2105.13889.pdf accepted in Neurips2021
Presenters
-
Aurélien Decelle
- Universidad Complutense de Madrid
Authors
-
Aurélien Decelle
- Universidad Complutense de Madrid
-
Beatriz Seoane
- Universidad Complutense de Madrid
- Univ Complutense
-
Cyril Furtlehner
- Paris Saclay University
- Inria, Université Paris Saclay
-
-
Long range memory in deep neural networks' neural activations
ORAL
–
Presenters
-
Ling Feng
- Natl Univ of Singapore
Authors
-
Ling Feng
- Natl Univ of Singapore
-
Nicholas Jia Le Chong
- National University of Singapore
-
-
Identifying symmetries in the statistical ensemble of coarse-graining rules
ORAL
–
Publication: arXiv:2103.16887, arXiv:2101.11633
Presenters
-
Doruk Efe Gokmen
- ETH Zurich
Authors
-
Doruk Efe Gokmen
- ETH Zurich
-
Zohar Ringel
- The Hebrew University of Jerusalem
-
Sebastian Huber
- ETH Zurich
-
Maciej Koch-Janusz
- Univ of Zurich
-
-
Exploring the loss landscape with Langevin dynamics
ORAL
–
Presenters
-
Théo Jules
- Raymond and Beverly Sackler School of Physics and Astronomy
- Tel Aviv University
Authors
-
Théo Jules
- Raymond and Beverly Sackler School of Physics and Astronomy
- Tel Aviv University
-
Yohai Bar-Sinai
- Google LLC
- Tel Aviv University
-
-
Learning actions from data using invertible neural networks
ORAL
–
Presenters
-
Claudia Merger
- RWTH Aachen University
Authors
-
Claudia Merger
- RWTH Aachen University
-
Carsten Honerkamp
- RWTH Aachen University
-
Alexandre René
- RWTH Aachen University and University of Ottawa
-
Moritz Helias
- Institute of Neuroscience and Medicine (INM-6) and Institute for Advanced Simulation (IAS-6) and JARA BRAIN Institute I, Jülich Research Centre and RWTH Aachen University
-