A 2 year postdoctoral position shall be opened in ENS
in September 2018, funded by the ANR-FNS French-Swiss grant PAIL (PIs : Florent Krzakala
and Guilhem Semerjian, ENS-Paris and Nicolas Macris EPFL-Lausanne).
This project on Phase diagrams and Algorithms for Inference and Learning aims at developing a better theoretical understanding of the recent advances in machine learning and high-dimensional statistics that have had a tremendous impact but call for more principled, if not rigorous, justifications. This goal will be pursued in this project by building on the fruitful interplay between statistical mechanics of disordered systems, discrete mathematics and computer science that has produced a number of important results in the related fields of constraint satisfaction problems and probabilistic inference.
Keywords: Machine Learning, Signal processing,
information theory, Probability theoery, graphical models, Bayesian inference, compressed
sensing, error correcting codes, spatial coupling, Belief
Propagation, Message Passing, Deep Learning. Statistical physics -
glasses, spin glasses, random optimization problems, cavity method,
replica method - c, c++, matlab, julia, python.
Deadline for applying: 31st March 2018.
To apply, and for further
candidates should send their detailed cv (including list of
publication, presentations, citations etc.), and 1 page letter of
motivation explaining why they want to work on this subject, what
is their related experience, and present a short
project. Preselected candidates should be ready to provide two
letters of recommendation at a later stage.