Joseph Salmon

Current position

Assistant Professor at TELECOM ParisTech
Associate member at INRIA Parietal
Visiting assistant professor at UW, Statistics departement

Contact

TELECOM ParisTech,
46, rue Barrault,
75634 Paris Cedex 13, France
Office: E 410
Phone: +33 1 45 81 75 47
Email: joseph.salmon aatt telecom-paristech.fr
Github: https://github.com/josephsalmon
Twitter : https://twitter.com/salmonjsph

News

I'll be a visiting assistant professor at UW, statistics departement until mi-August.
celer: new and fast Lasso solver: code (github) and companion paper called "Dual Extrapolation for Faster Lasso Solvers" (pdf).
Joint work with M. Massias and A. Gramfort.
Our paper on Heteroscedastic Concomitant Lasso accepted to AISTATS2018.
HDR defense: June 29, ENS Paris-Saclay (manuscript, slides).
classgrade: an open source project for peer grading, mostly developped by C. Marini. In the process of being merged into: Peeramid
MOOC: Fondamentaux pour le Big Data (in French)

Latest papers

	On the benefits of output sparsity for multi-label classification E. Chzhen, C. Denis, M. Hebiri, J. Salmon, 2017. [bib] \| [pdf]
	A hierarchical Bayesian perspective on majorization-minimization for non-convex sparse regression: application to M/EEG source imaging Y. Bekhti, F. Lucka, J. Salmon, A.Gramfort, 2017. [bib] \| [pdf]

	A sharp oracle inequality for Graph-Slope P. C. Bellec, J. Salmon, S. Vaiter, Electron. J. Statist., vol.11, pp. 4851--4870, 2017. [Python Code] \| [bib] \| [pdf]
	Gap Safe screening rules for sparsity enforcing penalties E. Ndiaye, O. Fercoq, A. Gramfort, J. Salmon, JMLR, vol.18, pp. 1-33, 2017. [bib] \| [pdf]

	Dual Extrapolation for Faster Lasso Solvers M. Massias, A. Gramfort, J. Salmon, ICML, 2018. [Python Code] \| [bib] \| [pdf]
	Generalized Concomitant Multi-Task Lasso for sparse multimodal regression M. Massias, O. Fercoq, A. Gramfort, J. Salmon, AISTATS, 2018. [Python Code] \| [bib] \| [pdf]

Resume

From 2011 to the end of 2012, I was a post-doctoral Associate at Duke university working with Rebecca Willett in the Network and Imaging Science Laboratory (NISLab). In 2010, I finished my PHD in Statistics under the supervision of Dominique Picard and Erwan Le Pennec at the Laboratoire de Probabilités et de Modélisation Aléatoire (LPMA ) in Université Paris Diderot.
My thesis was about the denoising of digital images combining patch-based method and statistical aggregation. The full manuscript can be found in the section papers. I am especially interested in topics such as: high dimensional regression, aggregation of estimators, patch based denoising methods and non-parametric regression.

Research

You can find my papers and my talks slides. I also collect here software and other numerical elements for reproducing the experimental results presented in my papers.

Teaching

TELECOM ParisTech (2013-)

Measure and integration
Machine learning
Probability theory (advanced)
Statistics
Continuing education

ENS Lyon (2013-2015)

Master 2: Machine Learning; with Zaïd Harchaoui, Laurent Jacob and Julien Mairal
Course website

Paris 7 - Diderot (2013-)

Master 2: Statistical learning; with Stéphan Clémençon
Course website

ATER, Paris 7 - Diderot (2010-2011)

In 2010-2011, I was a lecturer in descriptive statistics for first and second year students with Stéphane Boucheron and A. Belaiche. I was also a lecturer in Statitics at ENSAE for second year students (Master 1 level).

Monitorat, Paris 7 - Diderot (2007-2010)

In 2009-2010, I supervised practical works in Econometrics, for student with level Licence 3. Releavant exercices are available here. I was also a lecturer in Statitics at ENSAE for second year students (Master 1 level).
In 2008-2009, I was a project supervisor in mathematics and algorithmic (Master 1 level) with Pierre Alquier and Mohamed Hebiri. Courses, project subjects and exercice sheets are available here.
In 2007-2008, I was a lecturer in Formal Calculus with Maple (Licence 1). Courses and exercices sheets are available in french here.

Other

Here you can find my resume and several links about mathematics, my colleagues and various stuff.