Some of my past presentations (most of the recent ones are videos):
Applied Machine Learning Days, EPFL, AI & Physics track, January 2020
Statistical physics for machine learning.
NeurIPS 2019 contributions, December 2019: 2 invited talks as
workshop and a pannel I moderated
Understanding Machine Learning via Exactly Solvable Statistical Physics Models;
The spiked matrix model with generative priors;
Theory Panel Discussion.
IPAM, UCLA, November 2019
Understanding machine learning via exactly solvable statistical
physics models.
MIFODS  Stochastics and Statistics Seminar, MIT, November 2019
Understanding ML with statistical physics.
MLAI in Telecom Paris October 2019
When statistical physics meets machine learning.
Workshop on Science of Data Science, ICTP, Trieste, October 2019
Analysing the gradient descentbased dynamics in complex landscapes.
40 years of Replica Symmetry Breaking, September 2019, Rome
Analyzing performance of gradientbased algorithms.
CIRM, Marseille, June 2019
Algorithms in highdimensional nonconvex landscapes.
ICML 2019 workshop Theoretical Physics for Deep Learning, LA, June 2019
Loss landscape and behaviour of algorithms in the spiked matrixtensor model.
Academie des sciences, Conference : De la physique
statistique a lâ€™intelligence artificielle, June 2019
Statistical physics modelling of machine learning (in French).
MATH + X Symposium, Rice University, Houston, January 2019
Computational threshold in simple models on neural networks.
Institut des Hautes Etudes Scientifiques (IHES) for the Itzykson
seminar, 21st November 2018, Phase transitions in regression and simple neural networks.
A short talk to large public at the conference France is AI'18
in Paris
Statistical Physics Modelling of Machine Learning.
Talk I gave in 50emes Journees de la Statistique 2018 in
Saclay on
Generalized Linear Regression: Optimal Errors and Algorithm for
Random Instances..
A talk including historial perspective and recent advances in Statistical Mechanics
of Learning a rule I gave in IPhT at the Claude Itzykson conference
celebrating Cyrano DeDominicis, June 2018,
Statistical Mechanics of Learning a rule.
Three lectures (3 times 1.5h) on Random Constraint Satisfaction and
Inference I gave in ICTS in Bangalore.
Seminar I gave in the Simons Institute for Theory of Computing in
Berkeley on
Solvable Model of Unsupervised Feature Learning.
Seminar I gave in Big Data conference in Harvard on Clustering in
Networks: Phase Transitions and optimal algorithms.
Seminar for rather large public on Statistical Physics of Complexity.
Colloquim I gave in the physics department of ENS on Clustering
of networks: From phase transitions to optimal algorithms.
Statistical Physics of
Infence: Thresholds and Algorithms;
presentation of my HDR thesis in November 2015.
A video of a colloquim I gave in the physics department of
ENS on Clustering
of networks: From phase transitions to optimal algorithms in
March 2015.
A video of a talk I gave for a rather large public on
Statistical Physics of Complexity. Third Session of the
Visionary & Retrospective Talks club, at the Center for Research
and Interdisciplinarity, Paris, January 2015
Phase transitions and sample complexity in
Bayesoptimal matrix factorization;
talk at GDR ISIS workshop in Paris, January, 2014.
Clustering in networks:
Phase transitions and optimal algorithms;
talk given at Journee de Physique Statistique in Paris, January, 2014 and several
other occasions.
From crystal nucleation
to fast data acquisition;
talk at STATPHYS 2013 in Seoul, July, 2013.
Nonadaptive pooling strategies for detection of rare faulty items;
talk at a ICC workshop in Budapest, June, 2013.
Inferring the origin of epidemic with dynamic message passing;
talk given at a workshop in Santa Fe, May, 2013.
Where the
really hard problems really are?;
review talk at a workshop in Tokyo, January, 2013.
Compressed Sensing;
Seminar in ENS, Paris, November 2012.
Phase transitions and algorithmic
barriers in inference problems; invited talk given at IPAM, UCLA, in
January 2012. Includes examples of community detection and
compressed sensing, physics explanation of why the spatial
coupling in compressed sensing saturates the threshold.
Smart Grid;
Introductory lecture about Smart Grid given in ENS, December 2011.
Random Adversarial Satisfiability Problem; invited talk in Bardonecchia, Italy, February 2011.
Inference of functional modules in networks; invited talk in Tokyo, Japan, November 2010.
Inference of parameters in particle tracking by passing messages between images; talk in Hong Kong, China, July 2010.
Glassy phases: A possible origin of computational hardness; invited talk on the March Meeting in Portland, March 2010.
Integration of renewables in redundant power grid; presented at HICSS43, Hawaii, January 2010.
Random Field Ising Model; no spin glass phase in random field Ising model, RFIM at fixed magnetization, conjecture about graph partitioning. Presented at Banff, Canada, November 2009.
Cavity method for slow dynamics; generalization of the cavity method for adiabatic evolution of glassy Gibbs states; presented at LPT University ParisSud, October 2009.
Learning from and about complex energy landscapes; presented as a Santa Fe Institute Colloquium, September 2009.
Physics on random graphs; presented at Physics of Computations seminar, University of Chicago, September 2009.
Statistical physics of hard optimization problems; defense of my PhD thesis, June 2008 in University ParisSud.
