{"id":548040,"date":"2018-11-14T12:46:29","date_gmt":"2018-11-14T20:46:29","guid":{"rendered":"https:\/\/www.microsoft.com\/en-us\/research\/?post_type=msr-event&p=548040"},"modified":"2025-08-06T11:56:43","modified_gmt":"2025-08-06T18:56:43","slug":"physics-ml-workshop","status":"publish","type":"msr-event","link":"https:\/\/www.microsoft.com\/en-us\/research\/event\/physics-ml-workshop\/","title":{"rendered":"Physics \u2229 ML"},"content":{"rendered":"\n\n

Venue: <\/strong>
\nMicrosoft, Building 99,
\nRedmond, Washington, USA<\/a><\/p>\n

Watch on demand:<\/strong>
\nDay 1: 9:00 AM\u201310:30 AM (opens in new tab)<\/span><\/a>
\nDay 1: 11:00 AM\u201312:30 PM<\/a>
\nDay 1: 2:00 PM\u20134:05 PM<\/a>
\nDay 2: 9:00 AM\u20139:45 AM<\/a>
\nDay 2: 10:15 AM\u201312:30 PM<\/a>Opens in a new tab<\/span><\/p>\n

\n

Update May 2020:<\/strong> Following up on our workshop here, we have begun a virtual biweekly seminar series continuing our conversations on the interface between physics and ML. Please visit http:\/\/physicsmeetsml.org\/ (opens in new tab)<\/span><\/a> for more information.<\/p><\/blockquote>\n<\/div>\n

<\/div>\n

The goal of Physics<\/em> \u2229 ML<\/em> (read ‘Physics Meets ML’) is to bring together researchers from machine learning and physics to learn from each other and push research forward together. In this inaugural edition, we will especially highlight some amazing progress made in string theory with machine learning and in the understanding of deep learning from a physical angle.\u00a0Nevertheless, we invite a cast with wide ranging expertise in order to spark new ideas.\u00a0Plenary sessions from experts in each field and shorter specialized talks will introduce existing research.\u00a0We will hold moderated discussions and breakout groups in which participants can identify problems and hopefully begin new collaborations in both directions. For example, physical insights can motivate advanced algorithms in machine learning, and analysis of geometric and topological datasets with machine learning can yield critical new insights in fundamental physics.<\/p>\n

Organizers<\/h3>\n

Greg Yang<\/a>, Microsoft Research
\nJim Halverson (opens in new tab)<\/span><\/a>, Northeastern University
\nSven Krippendorf, LMU Munich
\nFabian Ruehle, CERN, Oxford University
\nRak-Kyeong Seong (opens in new tab)<\/span><\/a>, Samsung SDS
\nGary Shiu, University of Wisconsin<\/p>\n

Microsoft Advisers<\/h3>\n

Chris Bishop<\/a>, Microsoft Research
\nJennifer Chayes<\/a>, Microsoft Research
\nMichael Freedman<\/a>, Microsoft Research
\nPaul Smolensky<\/a>, Microsoft ResearchOpens in a new tab<\/span><\/p>\n

Day 1 | Thursday, April 25<\/h2>\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n
Time (PDT)<\/strong><\/td>\nSession<\/strong><\/td>\nSpeaker<\/strong><\/td>\n<\/td>\n<\/tr>\n
<\/td>\nSession 1<\/strong><\/td>\nPlenary talks<\/strong><\/td>\n<\/td>\n<\/tr>\n
8:00 AM\u20139:00 AM<\/td>\nBreakfast<\/td>\n<\/td>\n<\/td>\n<\/tr>\n
9:00 AM\u20139:45 AM<\/td>\nGauge equivariant convolutional networks<\/td>\nTaco Cohen<\/td>\n<\/td>\n<\/tr>\n
9:45 AM\u201310:30 AM<\/td>\nUnderstanding overparameterized neural networks<\/td>\nJascha Sohl-Dickstein<\/td>\n<\/td>\n<\/tr>\n
10:30 AM\u201311:00 AM<\/td>\nBreak<\/td>\n<\/td>\n<\/td>\n<\/tr>\n
11:00 AM\u201311:45 AM<\/td>\nMathematical landscapes and string theory<\/td>\nMike Douglas<\/td>\n<\/td>\n<\/tr>\n
11:45 AM\u201312:30 PM<\/td>\nHolography, matter and deep learning<\/td>\nKoji Hashimoto<\/td>\n<\/td>\n<\/tr>\n
12:30 PM\u20132:00 PM<\/td>\nLunch<\/td>\n<\/td>\n<\/td>\n<\/tr>\n
2:00 PM\u20134:05 PM<\/td>\nSession 2<\/strong><\/td>\nApplying physical insights to ML<\/strong><\/td>\n<\/td>\n<\/tr>\n
2:00 PM\u20132:45 PM<\/td>\nPlenary: A picture of the energy landscape of deep neural networks<\/td>\nPratik Chaudhari<\/td>\n<\/td>\n<\/tr>\n
2:45 PM\u20134:05 PM<\/td>\nShort talks<\/td>\n<\/td>\n<\/td>\n<\/tr>\n
<\/td>\nNeural tangent kernel and the dynamics of large neural nets<\/td>\nClement Hongler<\/td>\n<\/td>\n<\/tr>\n
<\/td>\nOn the global convergence of gradient descent for over-parameterized models using optimal transport<\/td>\nL\u00e9na\u00efc Chizat<\/td>\n<\/td>\n<\/tr>\n
<\/td>\nPathological spectrum of the Fisher information matrix in deep neural networks<\/td>\nRyo Karakida<\/td>\n<\/td>\n<\/tr>\n
<\/td>\nQ&A<\/td>\n<\/td>\n<\/td>\n<\/tr>\n
<\/td>\nFluctuation-dissipation relation for stochastic gradient descent<\/td>\nSho Yaida<\/td>\n<\/td>\n<\/tr>\n
<\/td>\nFrom optimization algorithms to continuous dynamical systems and back<\/td>\nRene Vidal<\/td>\n<\/td>\n<\/tr>\n
<\/td>\nThe effect of network width on stochastic gradient descent and generalization<\/td>\nDaniel Park<\/td>\n<\/td>\n<\/tr>\n
<\/td>\nQ&A<\/td>\n<\/td>\n<\/td>\n<\/tr>\n
<\/td>\nShort certificates for symmetric graph density inequalities<\/td>\nRekha Thomas<\/td>\n<\/td>\n<\/tr>\n
<\/td>\nGeometric representation learning in hyperbolic space<\/td>\nMaximilian Nickel<\/td>\n<\/td>\n<\/tr>\n
<\/td>\nThe fundamental equations of MNIST<\/td>\nCedric Beny<\/td>\n<\/td>\n<\/tr>\n
<\/td>\nQ&A<\/td>\n<\/td>\n<\/td>\n<\/tr>\n
<\/td>\nQuantum states and Lyapunov functions reshape universal grammar<\/td>\nPaul Smolensky<\/td>\n<\/td>\n<\/tr>\n
<\/td>\nMulti-scale deep generative networks for Bayesian inverse problems<\/td>\nPengchuan Zhang<\/td>\n<\/td>\n<\/tr>\n
<\/td>\nVariational quantum classifiers in the context of quantum machine learning<\/td>\nAlex Bocharov<\/td>\n<\/td>\n<\/tr>\n
<\/td>\nQ&A<\/td>\n<\/td>\n<\/td>\n<\/tr>\n
4:05 PM\u20134:30 PM<\/td>\nBreak<\/td>\n<\/td>\n<\/td>\n<\/tr>\n
4:30 PM\u20135:30 PM<\/td>\nThe intersect \u2229<\/td>\n<\/td>\n<\/td>\n<\/tr>\n<\/tbody>\n<\/table>\n

Day 2 | Friday, April 26<\/h2>\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n
Time (PDT)<\/strong><\/td>\nSession<\/strong><\/td>\nSpeaker<\/strong><\/td>\n<\/td>\n<\/tr>\n
<\/td>\nSession 3<\/strong><\/td>\nApplying ML to physics<\/strong><\/td>\n<\/td>\n<\/tr>\n
8:00 AM\u20139:00 AM<\/td>\nBreakfast<\/td>\n<\/td>\n<\/td>\n<\/tr>\n
9:00 AM\u20139:45 AM<\/td>\nPlenary: Combinatorial Cosmology<\/td>\nLiam McAllister<\/td>\n<\/td>\n<\/tr>\n
9:45 AM\u201310:15 AM<\/td>\nBreak<\/td>\n<\/td>\n<\/td>\n<\/tr>\n
10:15 AM\u201311:35 AM<\/td>\nShort talks<\/td>\n<\/td>\n<\/td>\n<\/tr>\n
<\/td>\nBypassing expensive steps in computational geometry<\/td>\nYang-Hui He<\/td>\n<\/td>\n<\/tr>\n
<\/td>\nLearning string theory at Large N<\/td>\nCody Long<\/td>\n<\/td>\n<\/tr>\n
<\/td>\nTraining machines to extrapolate reliably over astronomical scales<\/td>\nBrent Nelson<\/td>\n<\/td>\n<\/tr>\n
<\/td>\nQ&A<\/td>\n<\/td>\n<\/td>\n<\/tr>\n
<\/td>\nBreaking the tunnel vision with ML<\/td>\nSergei Gukov<\/td>\n<\/td>\n<\/tr>\n
<\/td>\nCan machine learning give us new theoretical insights in physics and math?<\/td>\nWashington Taylor<\/td>\n<\/td>\n<\/tr>\n
<\/td>\nBrief overview of machine learning holography<\/td>\nYi-Zhuang You<\/td>\n<\/td>\n<\/tr>\n
<\/td>\nQ&A<\/td>\n<\/td>\n<\/td>\n<\/tr>\n
<\/td>\nApplications of persistent homology to physics<\/td>\nAlex Cole<\/td>\n<\/td>\n<\/tr>\n
<\/td>\nSeeking a connection between the string landscape and particle physics<\/td>\nPatrick Vaudrevange<\/td>\n<\/td>\n<\/tr>\n
<\/td>\nPBs^-1 to science: novel approaches on real-time processing from LHCb at CERN<\/td>\nThemis Bowcock<\/td>\n<\/td>\n<\/tr>\n
<\/td>\nQ&A<\/td>\n<\/td>\n<\/td>\n<\/tr>\n
<\/td>\nFrom non-parametric to parametric: manifold coordinates with physical meaning<\/td>\nMarina Meila<\/td>\n<\/td>\n<\/tr>\n
<\/td>\nMachine learning in quantum many-body physics: A blitz<\/td>\nYichen Huang<\/td>\n<\/td>\n<\/tr>\n
<\/td>\nKnot Machine Learning<\/td>\nVishnu Jejjala<\/td>\n<\/td>\n<\/tr>\n
11:35 AM\u201312:30 PM<\/td>\nPanel discussion with panelists Michael Freedman, Clement Hongler, Gary Shiu, Paul Smolensky, Washington Taylor<\/td>\n<\/td>\n<\/td>\n<\/tr>\n
12:30 PM\u20131:30 PM<\/td>\nLunch<\/td>\n<\/td>\n<\/td>\n<\/tr>\n
<\/td>\nSession 4<\/strong><\/td>\nBreakout groups<\/strong><\/td>\n<\/td>\n<\/tr>\n
1:30 PM\u20133:00 PM<\/td>\nPhysics breakout groups<\/td>\n<\/td>\n<\/td>\n<\/tr>\n
<\/td>\nSymmetries and their realisations in string theory<\/td>\nSergei Gukov, Yang-Hui He<\/td>\n<\/td>\n<\/tr>\n
<\/td>\nString landscape<\/td>\nMichael Douglas, Liam McAllister<\/td>\n<\/td>\n<\/tr>\n
<\/td>\nConnections of holography and ML<\/td>\nKoji Hashimoto, Yi-Zhuang You<\/td>\n<\/td>\n<\/tr>\n
3:00 PM\u20134:30 PM<\/td>\nML breakout groups<\/td>\n<\/td>\n<\/td>\n<\/tr>\n
<\/td>\nGeometric representations in deep learning<\/td>\nMaximilian Nickel<\/td>\n<\/td>\n<\/tr>\n
<\/td>\nUnderstanding deep learning<\/td>\nYasaman Bahri, Boris Hanin, Jaehoon Lee<\/td>\n<\/td>\n<\/tr>\n
<\/td>\nPhysics and optimization<\/td>\nRene Vidal<\/td>\n<\/td>\n<\/tr>\n<\/tbody>\n<\/table>\n
<\/div>\n

Opens in a new tab<\/span><\/p>\n

\t

\n\t\t\t\t\t
\n\t\t\t\t\n\t\t\t\t\t\n\t\t\t\t\t\tExpand all\t\t\t\t\t<\/button>\n\t\t\t\t\t | <\/span>\n\t\t\t\t\t\n\t\t\t\t\t\tCollapse all\t\t\t\t\t<\/button>\n\t\t\t\t<\/div>\n\t\t\t<\/div>\n\t\t\t\t