Master 2

Mathematics,Vision Learning


Overview


There has been a massive expansion in the use of numerical data in all areas of science, technology and society, creating a need to train top-level researchers in mathematics, with expertise in numerical data acquisition and processing, as well as its automatic interpretation. These two aspects are fully complementary and are reflected in the three terms characterizing the MVL degree program: M for mathematics, V for vision, L for learning.

The courses offered are thus driven by data and problems from the real world: scientific fields, industrial and medical applications. Many mathematical topics are covered: signal representation techniques, variational methods and partial differential equations in image analysis, compressed sensing, probability learning theory, random matrices, convex optimization, theory of shape space, kernel learning methods, graphic models, Markovian simulation learning, control theory, reinforcement learning, etc.

Language of instruction: English and French
ECTS: 60 (lectures 8×5 ECTS, traineeship 20 ECTS)
Oriented: Industry and research
Duration: 1 year
Courses Location: mostly ENS Paris Saclay, also Telecom ParisTech, ENS Ulm, Polytechnique, Centrale.


Educational objectives


The MVL degree program enables students who are mathematicians, computer scientists, engineers, and physicists by training to discover a range of concepts, models, and algorithms. This will enable them to undertake research topics at the numerical interfaces of mathematics. By rooting most of these courses in applied fields, students learn about all aspects of an applied research project, right up to validating methodologies and algorithms via numerical experiments on real data.


Program structure


Courses take place at one of two periods dividing the first semester (respectively october-december and January-March). They are organised in two category : « data/modeling » (related to a specific type of data) and « learning » (dedicated to general learning methodology). Students should validate 8 lectures (for a total of 40 ECTS), with a minimum of one of each category at each period. Lectures are validated by written exams, oral exams or applied projects, depending on the nature of the course.

First period – Data/Modeling

  • Topological data analysis for imaging and machine learning
    J. TIERNY, F.CHAZAL
  • Introduction to medical image analysis
    H. DELINGETTE, X. PENNEC
  • Introduction à l’imagerie numérique
    J. DELON,Y. GOUSSEAU
  • Object recognition and computer vision
    I. LAPTEV, J. PONCE, C. SCHMID, J. SIVIC,
  • Sub-pixel image processing
    L. MOISAN
  • 3D computer vision
    R. MARLET, P. MONASSE, M. AUBRY
  • Image denoising : the human machine competition
    J.-M. MOREL, G.FACCIOLO, P.ARIAS
  • Méthodes mathématiques pour les neurosciences
    E. TANRE, R. VELTZ

First period – Learning

  • Computational statistics
    S.ALLASSONNIERE
  • Convex optimization and applications in machine learning
    A. D’ASPREMONT
  • Probabilistic graphical models
    P.LATOUCHE, N. CHOPIN
  • Reinforcement learning
    A. LAZARIC
  • Computational optimal transport
    G. PEYRE
  • Graphs in machine learning
    M. VALKO
  • Introduction to statistical learning
    N. VAYATIS
  • Advanced learning for text and graph data ALTEGRAD
    M. VAZIRGIANNIS
  • Learning and Algorithmic Game Theory
    V. PERCHET
  • Deep Learning
    V.LEPETIT
  • Foundations of Distributed and Large Scale Computing Optimization
    E. CHOUZENOUX et J.-C. PESQUET

Second period – Data/Modeling

  • Remote sensing data: from sensor to large-scale geospatial data exploitation
    F. TUPIN, G. FACCIOLO, A.ALMANSA
  • Audio signal processing – Time-frequency analysis
    E. BACRY
  • Parcimonie et analyse de données massives en astrophysique
    J. BOBIN, J.-L. STARCK
  • Nuages de points et modélisation 3D
    T. BOUBEKEUR, J-E. DESCHAUD, F. GOULETTE,
  • Imagerie fonctionnelle cérébrale et interface cerveau machine
    M. CLERC, T. PAPADOPOULOS, B. THIRION
  • Deformable models and geodesic methods for image analysis
    L. COHEN, G. PEYRE
  • Méthodes mathématiques pour l’analyse d’images (The mathematics of Imaging)
    A. DESOLNEUX, B. GALERNE
  • Audio signal Analysis, Indexing and Transformations
    G. RICHARD, R.BADEAU
  • Géométrie et espaces de formes
    A. TROUVE, J. GLAUNES
  • Problèmes inverses et imagerie : approches statistiques et stochastiques
    J. GARNIER
  • Algorithms for speech and natural language processing
    E. DUPOUX, B. SAGOT
  • Biostatistics
    R. PORCHER

Second period – Learning

  • Théorie des matrices aléatoires et apprentissage
    R. COUILLET, J. NAJIM
  • Modélisation en neurosciences et ailleurs
    J.-P. NADAL
  • Kernel Methods for machine learning
    J. MAIRAL, J.-P. VERT
  • Approches géométriques en apprentissage statistique: l’exemple des données longitudinales
    S.DURRLEMAN
  • L’apprentissage par réseaux de neurones profonds
    S.MALLAT
  • Predictions of individual sequences
    P.GAILLARD
  • Fondements Théoriques du deep learning
    S.GERCHINOVITZ, F.MALGOUYRES, E.PAUWELS, N.THOME
  • Deep Learning in Practice
    G.CHARPIAT, E.OYALLON
  • Apprentissage Profond pour la Restauration et la Synthese d’Images
    A.ALMANSA, A.NEWSON, S.LADJAL
  • Reading group
    V.PERCHET

IP Paris labs involved



Career prospects


The main openings for MVL graduates are in applied research with big research bodies (CNRS, CEA, CNES, INRIA, INRA, INSERM, etc.) or in the R&D centres of big companies (SAFRAN, General Electric, Technicolor, Saint-Gobain, SAGEM, Dassault Systèmes, Xerox, etc.).

A large majority of students engage in a PhD program, either in an industrial or academic laboratory. In recent years, SMEs and innovative startups in the digital world, in France and abroad, are looking for the profiles of this training as well.


Institutional partners


  • Université Paris Descartes,
  • Centrale Supélec,
  • ENS Paris, (Université Paris-Saclay)
  • INRIA (Lille Nord Europe, Saclay IdF , Sophia Antipolis)
  • Ecole des Ponts ParisTech,
  • ENSTA Paris
  • Mines ParisTech
  • Université Paris Dauphine

Admissions


Application guidelines for a master’s program at IP Paris

Academic prerequisites

  • Completion of the 1st year of a master program (Master 1)
  • Typically, those of a training in applied mathematics at level M1 but profiles of motivated computer scientists of very good mathematical level are also considered.

Language Prerequisite

  • English and French

Application timeline

Deadlines for the Master application sessions are as follows:
– First session: February 28, 2020
– Second session: April 30, 2020
– Third Session (optional): June 30, 2020 (only if there are availabilities remaining after the 2 first sessions)
Applications not finalized for a session will automatically be carried over to the next session.

You shall receive an answer 2 months after the application deadline of the session.


Tuition fees


National Master: Official tuition fees of the Ministry of Higher Education, Research and innovation (2019-2020, EU students: 243 euros / Non-EU students: 3770 euros)


Contact


Yann Gousseau – Télécom Paris
Email
Agnès Desolneux – ENS Saclay
Email
Nicolas Vayatis – ENS Saclay
Email
Secrétariat : Delphine Laverne
Email