M1 Parisian Research Master in Computer Science (MPRI)

Voir description dans le parcours-type QDCS.

Arpad Rimmel, Marc-Antoine.

In this course, students will learn to appreciate optimization problems and determine the best approaches to deal with them.
The notions of complexity and polynomial reductions will be seen and practiced.
Some algorithmic solution techniques will be presented that solve difficult problems (approximation, approximation scheme, heuristics). This course will be based on the modeling of concrete problems.

A basic complexity course.

ORSAY

Weisser.

The objective of the course is to present the main structural results and algorithms on graphs, in connection with the concrete situations in which they are used. The following points will be studied:
- Graph coloring, number/chromatic index;
- Cycle and path problems, Eulerian paths, Hamiltonian cycles;
- Couplings (algorithms, structural results, couplings in bipartite graphs, minimum cover per vertex);
- Connectedness (edge/vertex connectedness, Menger theorem);
- Planar graphs;
- Theory of minors and their applications (FPT, tree decomposition).

Algorithms, graph algorithms, complexity.

ORSAY

Oguz Kaya.

FR: L'évaluation du cours est basée sur un examen final écrit. Plan du cours: -Description du modèle de machine PRAM. -Travail et profondeur des algorithmes PRAM, optimalité de travail, théorème de Brent. -Conception et analyse d'algorithmes parallèles utilisant le modèle PRAM. -Description des machines parallèles à mémoire distribuée, des réseaux de communication et des algorithmes de communication. -Algorithmes parallèles de graphes et d'arbres. -Algorithmes parallèles de tri et de recherche. -Algorithmes parallèles du calcul matriciel.

EN: Course grading is based on a final written exam. Course plan: -Description of the PRAM model. -Work and depth of PRAM algorithms, work optimality, Brent's theorem. -Design and analysis of parallel algorithms using the PRAM model. -Description of the distributed memory parallel machines, communication networks, and communication algorithms. -Parallel graph and tree algorithms. -Parallel sorting and searching. -Parallel linear algebra kernels.

The goal of this course is to provide an adequate theoretical background for the design and analysis of parallel algorithms in different parallel computing environments. The course starts with the introduction of an ideal parallel machine (parallel random-access-machine, or PRAM), then focuses on designing various optimal algorithms and analyzing their complexity in this setting. Then, distributed-memory parallel architectures with different communication networks are introduced. Finally, the parallelization of many well-known fundamental algorithms are discussed.

Basic algorithmic skills Programming knowledge Basic understanding of a computer architecture.

-Introduction to parallel computing (Grama, Gupta, Karypis, Kumar). -Parallel algorithms (Blelloch).

ORSAY - GIF-SUR-YVETTE

Joanna Tomasik, Johanne Cohen.

Objective: to develop skills in the area of discrete optimization.
The course presents algorithms that solve problems beyond class P and those problems belonging to class P whose exact solution is too costly in practice. The time efficiency is obtained through probabilities. On the other hand, the results are not necessarily exact. This course deals with the notions of adversary and searching the state space.
- Introduction to probability theory;
- Interpretation of the results of a randomized algorithm (statistical evaluation of the performance: central limit theorem, confidence interval);
- Optimization algorithms without performance guarantees, i.e. heuristic algorithms (the concept of the state space and the corresponding search: Hill Climbing, simulated annealing with its underlying Markov chain);
- Techniques for accelerating heuristics: evolutionary algorithms (genetic algorithms);
- Online algorithms: design of algorithms when the input is revealed as it is received (the secretary problem, the ski rental problem, etc.);
- Introduction to game theory (games, Nash equilibria, potential game, computation of equilibria, application of the concept of Nash equilibrium in machine learning in adversarial networks).

ORSAY

Kim Nguyen.

Le cours s'organise classiquement en séances de magistrales suivies de TD ou TP. L'évaluation se fait par contrôle continu (devoirs, interro, TP notés ou mini-projets) et par un examen écrit.

Learning objectives: to study the various forms of automata, including tree automata and their relationship to logics.
Description:
This course presents various extensions to word automata, classically used in syntactic analysis.
It focuses on tree automata and regular tree languages.
It also presents connections between logic and automata : the way in which the latter offers decision procedures for the former.
Finally, the course shows how automata can be used in practical cases e.g. for the sake of typing procedures, filtering compilation or validation of XML or JSON documents.
Topics covered:
- Recap on finite word automata and regular languages;
- Tree automata and regular tree languages;
- Logic, automata and their relations;
- Alternating automata;
- Constrained automata;
- Tree transducers.

Bachelor level knowledge of formal languages (grammars, word automata, regular expressions). Programming in a high-level language (OCaml, Java or Python).

ORSAY

Florent Hivert, Nicolas Thiéry.

This course is an introduction to mathematical tools for symbolic computation and combinatorics, with applications for example in cryptography cryptography or to software random testing. We will adopt an effective approach by focusing on both algorithms and their implementations in computing systems.

3rd year Bachelor Algorithms course, 2nd year Linear Algebra course, Maths for Computer Science.

ORSAY

Classes in space (and a bit more);
Notion of computational model, Turinguniversality, universal machines;
Simple randomized complexity classes, approximation classes & inclusions;
Reachability problems, P-completeness;
Semi-decidability, proofs by injection of universal computation.

A basic complexity course such as the "Theoretical Computer Science" course of the L3 Computer Science / Math-info of Paris Sud. Deterministic and non-deterministic Turing machines, P, NP and PSPACE classes, notions of reduction and completeness (at least NPcompleteness), undecidability (halting problem).

ORSAY

Pablo Arrighi.

Content: in this course, we will focus on understanding the quantum nature of information, through the multiple theorems that characterize it. We will illustrate them with applications in quantum cryptography & communications, quantum error correction and quantum simulation.
The following points will be presented:
postulates, no-cloning, no-distinguishability;
density matrices and quantum operations;
Dirac and Coecke notations;
superdense-coding protocols, teleportation, QRAC, BB84 key sharing...;
non-signalling vs. locality, Bell inequalities, certified randomness ;
state decomposition theorems and operators ;
fundamental theorem of quantum error correction ;
Hamiltonian quantum simulation ;
digital quantum simulation by quantum walks and quantum cellular automata.

All of the necessary reminders will be given. However, make sure you do like linear algebra: vectors, matrices, inner product, norm.

Benoît Valiron.

The objective of this course is to understand how
quantum algorithms work, analyzing their strengths
and limitations. First, we will introduce the
concepts behind quantum computation and study
the standard computational model: circuit families
and the quantum co-processor model. We will
discuss the structure of quantum algorithms and the
resulting programming model. We will then study
the most well-known quantum algorithms: Grover,
Shor, etc. Finally, we will discuss the medium-term
applications of quantum computation, in particular
in the context of machine learning.

Basic knowledge in algorithmic/programming and computer architecture.

The objective of this course is to introduce formal specification and deductive verification of programs through the use of Why3 and Frama-C tools.

Logic, programming.

ORSAY

Burkhart Wolff.

This course is an introduction to the theory and practice of Coq and Isabelle/HOL interactive proof systems.
Description:
- logical framework: simply typed lambda-calculus, higher order logic;
- modeling and proofs: algebraic types, inductive definitions, recursive definitions and termination, induction;
- proof tools: architecture of a proof assistant, elementary tactics.

Logic, recurrence.

ORSAY

Thibaut Balabonski.

Learning objectives : To learn the theoretical tools to formalize and understand the behaviour of programs, through the lambda-calculus in particular (a function-based computational model that lies at the core of functional programming languages) and its different semantics.
Description:
This course presents the lambda-calculus, a formalism describing the notion of a computable function in the same way as Turing machines do. It goes on to explain how this formalism and its variants can be used to formally describe the behavior of programs. We will present the notions of large-step semantics, expressed only in terms of the results obtained, and of small-step semantics, which detail the different execution steps. We enter the details of the reasoning techniques associated with these formalisms, i.e. the ways in which we can state and prove properties concerning the execution of a program or the correction of a transformation.
This course is related to the Compilation and Rewriting courses.

Logic, recurrence, general culture of programming languages.

ORSAY

Thibaut Balabonski.

Learning Objectives: Deconstruct the mechanisms of high-level programming languages in terms of elementary operations. Become familiar with symbolic manipulation of programs for interpretation, analysis, or translation purposes.
Description: This course addresses the core of the compilation process, namely the translation of high-level programs into more elementary programs. The starting point will be the structured representation of the program obtained after the syntactic analysis. The end point will be a directly executable program, for example in assembly language. The course also presents techniques for program analysis and optimization, and discusses the relevant aspects of architecture and systems.
Topics include:
- Control structures;
- Functions;
- Data types and structures
- Functional programming;
- Object-oriented programming;
- Memory management
- Register allocation.
The course leaves a large part to programming in the OCaml language. The practical classes incrementally lead to the construction of a compiler for a realistic language.

Syntactic analysis, general culture of programming languages, OCaml programming.

ORSAY

UE à choisir parmi l'offre de formation des parcours-types AI, ANO, DS, QDCS et M1 MPRI/ENS.

UE à choisir parmi l'offre de formation des parcours-types AI, ANO, DS, QDCS et M1 MPRI/ENS.

UE à choisir parmi l'offre de formation des parcours-types AI, ANO, DS, QDCS et M1 MPRI/ENS.

Antoine Manzanera Isabelle Bloch Henri Maître David Filliat.

2 or 3 reports on the practical sessions.

This course aims to give specific knowledge to the treatment of images for indexing and search by content:
extraction of image characteristics, descriptors, methods of classification and pattern recognition (with and without
learning), motion estimation, plane splitting, videos, indexing, search by content in image databases
and videos.

Machine Learning (TC0) Signal and Image (TC5).

Novembre - Décembre - Janvier.

ORSAY

Caio corro.

Recently, neural networks trained end-to-end have obtained impressive results in many problems related to natural languages (e.g. machine translation). These deep learning techniques do not rely on manual feature extraction or rule-based systems. However, behind the scenes a large part of this success is due to the development of neural architectures that are able to handle structured inputs and outputs.

In this course, we will study how build neural networks for problems related to natural languages. Specifically, we will learn how to:
- build dynamic computation graphs for sequential inputs,
- predict structures (e.g. text generation, semantic parsing),
- introduce inductive bias.
Moreover, we will develop a critical analysis of state-of-the-art NLP models:
- do they actually learn what we expect?
- how does deep for learning for NLP handle bias in training data?.

Machine Learning (TC 0) Deep Learning (OPT 4).

Novembre - Décembre - Janvier.

GIF-SUR-YVETTE

Anne Vilnat Pierre Zweigenbaum Sophie Rosset.

How can information be found in a text?
• Make a text in French understand a computer ... an old dream!
• Many applications, especially since the development of the web First material: the language
• recognize words, identify sentences, interpret texts • reminders and basic tools:
• a little grammar,
• analyzers
Extraction of information
• Switch from text to structured data
• detect concepts and relationships:
• what people or companies the press is talking about,
• what side effects of drugs are reported on
health forums
• methods based on large dictionaries
• learning methods
Concrete exploration of existing data in TD
Dialogue and understanding
Dialogue systems? conversational agents? chatbots?
• What is that ?
• How do we do ?
• approaches, models, data
• How are we evaluated?
• methodologies: objective versus subjective evaluations
• correlation?
• Focus on understanding
• understand what the user says and wants ? which models? Why ?.

Machine learning (TC0) Information retrieval (TC3) Probabilistic generative models (TC4).

Novembre - Décembre - Janvier.

GIF-SUR-YVETTE

Un grand nombre de TP sera donné en python pour illustrer les différentes notions acquises et regarder des applications.

A large number of computer lab. will be given in python in order the lecture and to develop some applications.

L'objectif du cours est de présenter les concepts et les résultats de la théorie de l'information à partir de l'analyse détaillée d'exemples issus de domaines applicatifs très divers.

The objective of this course is to present concepts and results of information theory starting from the detailed analysis of examples coming from very diverse fields of application.

Probability and statistics class.

Information Theory, Inference and Learning Algorithms, MacKay Elements of Information Theory, T. Cover.

Novembre - Décembre - Janvier.

GIF-SUR-YVETTE

0.4*contrôle continu + 0.6*examen.

The information society in which we live produces an ever-increasing flow of information of various types that needs to be dealt with quickly and effectively. Take, for example, a website among so many others: this site has a particular structure; it is composed of Web pages responding to a structuring of information that are their own and may vary. Thus, in the face of polymorphic data and knowledge, probabilistic models have emerged through their ability to grasp the variability of information. In the last decades, these models have become indispensable tools for information management and decision-making. The course is divided into three parts. The first part deals with structured and supervised learning while the second part deals with unsupervised models of contents. The third part will illustrate these first two parts by focusing on a specific application framework that is multilingual information processing.

TC4: Probabilistic generative models.

Novembre - Décembre - Janvier.

GIF-SUR-YVETTE

As deep learning technologies make impressive progress, it is tempting to think that the resolution of AI-complete problems such as Machine Translation is within reach. This class aims to provide the audience with an in-depth presentation of the domain of Machine Translation, emphasizing the difficulties that have been overcome, and the challenges ahead. Studying Machine Translation technologies is also of interest because this task serves as a model for many other applications of Natural Language Processing (NLP) and poses generic problems and challenges relating to modeling complex structured transduction phenomena in NLP (such as automatic grammar correction, text to speech, etc) from examples, or the algorithmic of searching very large structured output spaces. Machine Translation is useful for humans, but is also useful for automatic information processing systems, through transfert learring, and is one way to design multi-lingual processing component.

Organisation: This class will cover the following topics:
- History of Machine translation
- Alignment models, from sentence to word and subword alignment
- Modern architectures for Neural Machine Translation
- The challenges of Spoken Language translation
- Evaluating Machine Translation System
- Multi-lingual Machine Translation, Transfert Learning and Multi-lingual NLP.

Novembre - Décembre - Janvier.

GIF-SUR-YVETTE

Isabelle Bloch Antoine Manzanera.

0.4*contrôle continu + 0.6* synthèse d'article.

This course presents structural methods for image interpretation, with examples in medical imaging, remote sensing,
in robotics vision, in video (modeling and recognition activities, gestures, actions).
Knowledge-based methods, models, graphs, of spatial ontologies, grammars and information fusion will be studied.

TC5: Signal processing.

Novembre - Décembre - Janvier.

GIF-SUR-YVETTE

Michele Sebag.

0.4*contrôle continu + 0.6*projet (rapport, soutenance).

The objectives of this course are to understand and acquire practical experience with:
The definition of Reinforcement Learning (RL) problems, i.e. Markov Decision Processes.
Solution methods for model-based/model-free discrete RL problems.
Value function approximation for continuous RL problems.
Direct policy search with parameterized policies.
Applications of RL to robotic tasks.

Apart from the theoretical background provided in the lectures (cours magistraux), students will acquire hands-on experience by implementing a variety of discrete/continuous RL algorithms in Python during the lab hours (travaux pratiques).

Décembre - Janvier - Février.

GIF-SUR-YVETTE

Caio Corro Michèle Sebag.

50%: Exam 50%: Lab exercises.

This course covers:

introduction to neural networks,
efficient training of neural networks,
convolutional neural networks,
generative neural network (Variational auto-encoders, Generative auto-encoders, Adversarial auto-encoders).

Machine Learning (TC0).

Novembre - Décembre - Janvier.

GIF-SUR-YVETTE

Claude Barras Gael Richard Iona Vasilescu Laurence Devillers Matthieu Kowalski Chloé Clavel François Yvon.

Seminar.

Traitement du signal pour la parole
De la linguistique au traitement automatique de la parole.
Détection des émotions dans l'audio.
Separation de sources (cocktail party problem) et application au traitement de la parole
Deep pour traitement de la parole
Traitement automatique des langues.
Systèmes de traduction automatique.

TC1: Machine Learning TC4: Probabilistic Generative Models.

Décembre - Janvier - Février.

GIF-SUR-YVETTE

0.4*contrôle continu + 0.6*projet (rapport, soutenance).

1- Apprentissage à partir de flux de données
- Les défis
- le compromis réactivité - exploitation
- les approches

2- Apprentissage incrémental et constructif
- Apprentissage hiérarchique
- Notion de transfert dans le temps

3- Apprentissage multi-tâches et apprentissage par transfert entre domaines.
- Méthodes d'ensemble et Apprentissage collaboratif.

TC1: Machine Learning.

A. Cornuéjols, L. Miclet & V. Barra "Apprentissage artificiel. Deep learning, concepts et algorithmes", Eyrolles, 2018.

Décembre - Janvier - Février.

GIF-SUR-YVETTE

Dimo Brockhoff.

0.4*contrôle continu + 0.6*examen.

Black-box optimization is concerned with the optimization of difficult optimization problems, where the function to be optimized is seen as a black-box that can only return function values at given queried points. In particular, no analytical expression of the function is assumed, no knowledge about the function class can be exploited. Also the gradients are typically not available.

This advanced lecture aims at giving a broad overview of black-box algorithms for discrete and continuous problems. It will cover theoretical foundations as well as practical applications.

TC2: Optimization.

Décembre - Janvier - Février.

GIF-SUR-YVETTE

Novembre - Décembre - Janvier.

GIF-SUR-YVETTE

Alexandre Gramfort Thomas Moreau.

Half of your grade will come from the data challenges (performance on leaderboard) The other half will come from the team project.

You will put your basic machine learning and data analysis knowledge to test by solving practical data science problems in scientific or industrial applications
1) Entering a data challenge
2) Designing data science workflows.

PRE4: Scientific programming TC1: Machine Learning.

Novembre.

PALAISEAU

Ce cours est un cours de mise à niveau en statistique pour les étudiants de science des données et intelligence artificielle. Il inclura 7 séances de 3 heures, moitié cours, moitié TDs sur ordinateurs:
1- statistiques descriptives, distributions standard
2- estimation ponctuelle et par intervalle
3- estimation statistique fréquentiste vs Bayésienne (maximum de vraisemblance, maximum a posteriori)
4- loi des estimateurs du maximum de vraisemblance
5- tests d'hypothèse (T-test, chi-square, ANOVA)
6- méthode des moindres carrés, régression linéaire
7- distributions multivariées, analyse discriminante linéaire

This course is a pre-requisite for students of data science and artificial intelligence. In will include 7 sessions of 3 hours, half courses, half TDs:
1- descriptive statistics and well-known distributions
2- point and interval estimation
3- frequentist estimation vs Bayesian estimation (Maximum likelihood and maximum a posteriori)
4- distribution of maximum likelihood estimates
5- hypothesis tests (T-test, chi-square, ANOVA)
6- least squares method, linear regression
7- multivariate distributions, linear discriminant analysis.

Basic notions in probability and statistics.

Think Stats: Exploratory Data Analysis by Allen B. Downey Statistics in a Nutshell by Sarah Boslaugh All of statistics by Larry Wasserman.

Septembre.

GIF-SUR-YVETTE

Anne Auger Dimo Brockhoff.

(1) Linear algebra:
- linear map
- Vectors and matrices, vector-matrix operations
- Matrix inversions, diagonalization, determinants, eigen values
- Resolution of systems of linear equations
- Linear regression (multi-dimensional) with pseudo-inverse [if not taught in other classes]
- norm, scalar products, basic inequalities (triangular inequality)
- Linear transforms such as Fourier Transform and convolutions [if needed]

(2) Functional analysis
- Hilbert spaces [could be done quickly if needed]
- Limits, bounds, convergence, uniform convergence [how much is needed for other modules]

(3) Calculus:
- Derivatives and integrals
- Jacobian, Hessian matrices
- Linear operators (differential, integral)
- Simple differential and integral equations [if needed].

Septembre.

GIF-SUR-YVETTE

This class with cover basic knowledge about relational databases and SQL in preparation to TC6 (datacomp2).

Septembre.

GIF-SUR-YVETTE

Sylvain Caillou.

Basic and advanced Python
- Quick overview of Python Scientific libraries Ecosystem
- Development environment and tools
- Python Basics (types, containers, control flow, exceptions, context managers, ...)
- Advanced Python mecanism (packing/unpacking, comprehension, iterable/iterators, generators, ...)
- Good practices
- Design and packaging
Introduction to Numpy
- Numpy arrays
- Data Types
- Access element and manipulation
- Vectorization and Operations on arrays
Data exploration and preprocessing with Pandas
- Pandas data structures (DataFrame and Series)
- Data import and export
- Exploration
- Selecting and filtering
- Concatenating and merging
- Aggregating and grouping
- Working with missing values
- Time Series
Data visualization with Matplotlib, Seaborn and Bokeh
- Introduction to matplotlib (simple and advanced plots, customizing figure and axes, ....)
- Introduction to Seaborn and Bokeh
Introduction to Signal Processing with Scipy
- Spectral Analysis / Fourrier transforms
- Spectrogramms
- Frequency-domain filters
- Image filtering with convolution
Introduction to Scikit-Learn
- Representation of data
- Supervized learning (regression and classification)
- Unsupervized learning (clustering and dimensionality reduction)
- Model selection (cross validation, hyperparameters and model validation, model evaluation).

Good programming skills in at least one language Basic knowledge in Python.

Septembre.

GIF-SUR-YVETTE

1. Introduction to Machine learning

2. Classification with the perceptron

3. Linear models

4. Decision trees and random forests

5. Reinforcement learning

6. Dimensionality Reduction

7. Unsupervised learning.

[AI] PRE1: APPLIED STATISTICS [AI] PRE2: MATHEMATICS FOR DATA SCIENCE.

Septembre - Octobre.

GIF-SUR-YVETTE

Michele Sebag Francois Landes.

0.5*contrôle continu + 0.5*examen.

Goals:
The objective of this course is to provide the essential bases in machine learning or machine learning: the main families of models and the associated algorithms (inference and learning). In particular, will be treated:
- the theoretical bases of learning
- probabilistic models and Bayesian learning
- Linear models and SVM
content:
- General framework: representation, loss function, generalization and over-learning, entropy, supervised / unsupervised
- Bayesian frame and models
- Linear Models and Vector Machines Support

Practical part: projects and case studies based on Kaggle competitions.

David J. MacKay. Information Theory, Inference and Learning Algorithms. 2003. Cambridge University. (freely available) Richard O. Duda, Peter E. Hart, and David G. Stork. 2000. Pattern Classification (2nd Edition). Wiley-Interscience.

Septembre - Octobre.

GIF-SUR-YVETTE

Anne Auger Dimo Brockhoff.

0.4*contrôle continu+0.6*examen.

The objective of the course is to teach the basic theoretical and practical knowledge in continuous and discrete optimization.

In discrete optimization the lecture will cover complexity theory, dynamic programming, approximation algorithms and heuristics to linear programming. For continuous optimization, optimality conditions for unconstraint and constraint optimization, convex optimization will be covered. Gradient based, Newton, quasi-Newton and derivative-free algorithms will be presented.

PRE2: Mathematics for Data Science.

Convex Optimization, S. Boyd and L. Vandenberghe, Cambridge University Press. Introductory Lectures on Convex Optimization, Y. Nesterov, Springer.

Septembre - Octobre.

GIF-SUR-YVETTE

0.4*contrôle continu + 0.6*projet (rapport, soutenance).

In this module, we will discuss the main concepts and tools of information retrieval.
We will explore the different phases of the construction of the flagship application of this domain, the search engine: collection and cleaning of documents, detection of spam, indexing and implementation of information retrieval models. This will enable us to build a search engine that uses large volumes of information and responds to information needs in a relevant way.

Students will acquire basic methods for automatic language processing, retrieval and information retrieval.

Septembre - Octobre.

GIF-SUR-YVETTE

1 Non-probabilistic models
- K-means
- PCA
- (Weighted) Finite State Automaton

2. Probability recalls
- Motivation: non-deterministic events, modeling uncertainty, robustness to noisy data
- Random variables
- Divergence measure
- Generative versus discriminative models
- Minimum risk training and maximum likelihood estimation
- Monte-Carlo estimation

3. Simple Probabilistic Generative Models <- Pas de EM ici
- Multi-variate Gaussian
- Parzen Windows
- Gaussian mixtures
- Naive Bayes
- Discussion generative vs discriminative classifiers

4. Probabilistic Graphical Models
- Bayesian networks
- Conditionnal Independance and D-separation
- Causal Graphs and the œdo operator

5. Hidden Markov Models <-- dans ce cours on assume que les étiquettes sont observées dans les variables latentes, i.e. pas de EM
- Definition
- Parameter estimation
- Decoding: the Viterbi algorithm
- Marginalization in HMMs
- Discussion on discriminative graphical models (generative vs discriminative)

6. Latent variable models
- Expectation-Maximization
- Baum-Welch algorithm for HMMs with latent variables
- Semi-supervised learning with generative models

7. Deep Generative Models
- VAE
- GAN
- Normalizing Flows.

Septembre - Octobre.

GIF-SUR-YVETTE

Everything is online 1 typical session: - A brief introduction - Work on a small project (example: guitar tuner for spectral analysis) - A lot of questions.

Objectives:
- Basic techniques of signal processing: spectrum analysis, filtering
- Advanced techniques: Time-frequency, wavelets
- Inverse problems

Organization:
- Flipped classroom
- A lot of practice (Matlab or Python)

8 sessions
1 Intro + Spectral analysis (Guitar tuner + zoom in images)
2 Filtering (Delay effect FIR + IIR, image filtering)
3 Random signal and noise (Noise spectrum estimation and Wiener Filtering) .
4 Time-Frequency (Audio denoising).
5 Wavelets (Image denoising).
6 to 8 Project on inverse problem.

Basics of linear algebra and probabilities.

A Wavelet Tour of Signal Processing (S. Mallat) Numerical tour of Data Sciences: https://www.numerical-tours.com (G. Peyré).

Septembre - Octobre.

GIF-SUR-YVETTE

Project report.

To acquaint the students with algorithms, methods and techniques for the large scale matrices and their ubiquitous applications (text indexing, bigdata clustering, large social networks mining).

The students will have the following abilities after the course :
• be able to process a data set towards spectral analysis and find an optimal scheme for dimensionality reduction,
• be able to conduct feature selection and engineering
• have a good understanding on distributed learning for large scale data
• Well equipped to work in the industry that requires dominance of learning algorithms for large scale data
• Continue toward doctorate studies in relevant fields.

Competence that will be acquired:
Dealing with large scale distributed data
Distributed data management, MapReduce/Hadoop)
Distributed data mining and machine learning (dimensionality reduction, unsupervised learning, supervised learning)
Applications to bio data, social networks, and Web advertising.

PRE3: Datacomp 1.

Septembre - Octobre.

GIF-SUR-YVETTE

Description : Wireless networks and Internet of Things represent one of the major elements of our life in the next decades. Humans, Robots, Drones, vehicles and all type of things are going to be connected through wireless connections to a global Internet that we call the Internet of Things. The objectives is offer to students a complete and detailed landscape on the different protocols and technologies that will be used to realize the new system in which we are going to evolve.

ORSAY

Objectifs : savoir reconnaitre quel problème issu du domaine des réseaux doit être modélisé sous forme de graphe plutôt qu'une autre formulation mathématique.

Description : Les graphes font partie des modélisations importantes et possibles (autre que la programmation mathématique) concernant les problèmes de réseaux pour lesquels il s'agit non seulement d'évaluer un problème de routage (par exemple) mais également d'optimiser un critère tel que le cout, l'efficacité énergétique ou encore la qualité de service. Voici les notions qui seront abordées dans ce cours. Représentation des graphes. Cheminements et connexité. Arbres et arborescences. Parcours de graphes. Plus courts chemins. Arbres couvrants minimums. Réseaux de transports, telecom, de drones.

Compétences : modélisation des problèmes de réseaux à travers les graphes.

Éléments de base de la théorie des graphes, programmation de base.

ORSAY

Objectifs : savoir reconnaitre quel problème issu du domaine des réseaux doit être modélisé sous la forme d'un programme mathématique et savoir comment le résoudre.

Ce cours se déroule en deux étapes. La première consiste à faire un tour d'horizon des modèles (de programmation mathématique non linéaires en variables) les plus utilisés pour formuler des problèmes réels provenant des réseaux : programmes quadratiques en nombres entiers (convexes, non convexes), fractionnaires en variables mixtes, logarithmiques, arc tangente, bi-niveaux. La seconde étape propose d'étudier les méthodes de résolution associées aux modèles présentés dans un premier temps : reformulation d'un programme non linéaire sous la forme d'un programme linéaire qui peut être résolu via un solveur de programmation linéaire.

Compétences : modélisation des problèmes de réseaux à travers la programmation mathématique non linéaire en variables mixtes.

Éléments de base de la théorie des graphes, programmation de base.

ORSAY

Objectifs : savoir reconnaitre quel problème issu du domaine des réseaux qui doit être modélisé sous la forme d'un programme bi-objectif et savoir le résoudre.

L'optimisation multi-objectifs apparait quand on doit optimiser simultanément plusieurs objectifs contradictoires, ce qui amène à choisir une solution de compromis parmi une multitude de solutions possibles. Ce cours propose dans un premier temps les principes de l'optimisation multi-objectifs en décrivant les méthodes permettant de résoudre ce type de problème, et ce, en se basant sur des études de cas réels. Dans un second temps, il sera intéressant de voir comment évaluer les performances de ces méthodes et choisir la méthode la mieux adaptée à un problème donné.

Compétences : modélisation des problèmes de réseaux comportant plusieurs critères.

éléments de base de la programmation mathématique, programmation de base.

ORSAY

Marc Baboulin.

The focus of this course is the parallel programming in distributed memory systems, with the message-passing paradigm using the MPI library.
Fundamentals of MPI programming model, spawning multiple processes, process rank, communicators and sub-communicators, and the distributed memory concept
Point-to-point MPI communications
Blocking vs. non-blocking MPI communications
Collective communication algorithms and MPI routines
Barriers, synchronizations, deadlock prevention
Performance optimization/tuning, load balancing, communication cost
Hybrid parallelism MPI+OpenMP
Advanced MPI topics time permitting (MPI shared memory, neighborhood collectives, ...).

Knowledge of algorithms and programming in C/C++ Basics of computer architecture [M1 QDCS] Parallel algorithms (recommanded).

Introduction aux réseaux Ad-hoc sans fil sous GNU/Linux et au développement sur systèmes embarqués
Description. Ce module forme les étudiants à la programmation système et réseau dans un environnement TCP/IP pour l'embarqué. Les étudiants y apprennent la manipulation d'un système GNU/Linux pour configurer et utiliser un réseau WiFi en mode Ad-hoc, la compilation croisée (cross-compilation) pour le développement d'applications sur systèmes embarqués, ainsi que la mise en pratique d'algorithmes de la théorie des graphes pour optimiser l'utilisation d'un réseau dont la topologie est connue. À la suite de ce module, les étudiants auront acquis des compétences rares et très recherchés sur le marché professionnel des réseaux et systèmes embarqués.

ORSAY

Description : Ce cours sera dédié à l'étude de plusieurs notions essentielles des nouvelles générations de réseaux telles que la qualité de service (QoS), la mobilité (MIP) ou encore la signalisation dans IP (SIP, VoIP). Une série de travaux dirigés et de travaux pratiques permettront d'assimiler et de mettre en application les différentes notions étudiées.

ORSAY

Cette unité d'enseignement a pour objectif de présenter plusieurs notions avancées des réseaux informatiques et de télécommunications. Elle sera consacrée aux fondements et concepts de base des réseaux cellulaires : concept cellulaire, techniques d'accès au support radio (FDMA, TDMA, CDMA, CSMA-CA), en détaillant le fonctionnement de quelques systèmes représentatifs tels que la 3G, la 4G et la 5G.

ORSAY

Ce cours aborde la spécification et le test de systèmes distribués de protocoles de communication. Après des rappels sur les automates et les types abstraits de données, le cours présente une algèbre de processus et le langage SDL. On illustre leur utilisation sur des protocoles réalistes et réels. Ces spécifications permettent la dérivation de cas de tests. Nous aborderons les problématiques de génération de tests d'architecture de tests. Nous verrons comment générer des tests dans le format TTCN3 et enfin nous aborderons le test à partir de traces d'exécution à savoir le test passif.

ORSAY

Objectifs : savoir reconnaitre quel problème issu du domaine des réseaux doit être traité au moyen de la théorie des jeux.

La théorie des jeux est largement employée dans le domaine des réseaux, ce domaine proposant de nombreuses techniques de modélisation et de résolution des problèmes liés aux comportement des utilisateurs ou des fournisseurs dans les réseaux. Les éléments étudiés dans ce cours en lien avec des applications réseaux telles que le routage par exemple sont les suivants : jeux coopératifs, non coopératifs ; jeux sous forme stratégique (dominance, Nash, Stackelberg) ; jeux dynamiques.

Compétences : Savoir se servir de la théorie des jeux pour étudier le comprtement des usagers d'un réseau.

éléments de base de la théorie des jeux, programmation de base.

ORSAY

The course is organized in lecture classes and practical labs. The practical implementation of the concepts taught in the lectures will be applied during practical labs and assigned projects.

The course focuses on the algorithms involved in data related tasks, collectively grouped under the concept of "data mining". Data mining is a set of algorithms for transforming, modelling, and interpreting data that can be directly applied to Data Science tasks, or can be necessary as pre-processing step, before the data can be presented to an, e.g., machine learning task.

Lectures will cover a wide range of concepts such as:

- data mining algorithms: finding similar items (e.g., LSH), finding frequent items, dimensionality reduction techniques
- applications to advertising and recommendation on the Web, such as collaborative filtering
- data stream mining: applying data mining algorithms for cases where the data is presented at a fast pace, and memory is insufficient to store (or access) the entirety of the dataset.

Algorithms, Programming (Python/C/Java).

1. J. Leskovec, A. Rajaraman, J. Ullman. "Mining of Massive Datasets." Cambridge University Press 2. S. Abiteboul, I. Manolescu, P. Rigaux, M.-C. Rousset, P. Senellart. "Web Data Management." Cambridge University Press.

Septembre - Octobre - Novembre - Décembre.

ORSAY

Les travaux dirigés permettent de mettre en oeuvre les techniques étudiées en cours. Des lectures seront proposées aux étudiants pour compléter les interventions. Les travaux pratiques permettent d'expérimenter les différents outils mis à disposition par les systèmes pour scruter et agir sur les paramètres liés à la performance des applications.

L'objectif de ce cours est de dévoiler les mécanismes internes d'un SGBD (représentation physique, évaluation de requêtes). L'accent est mis sur le moteur de requêtes des SGBDs et l'optimisation des applications. Ces connaissances sont nécessaires autant aux développeurs d'application qu'aux administrateurs de base de données, par exemple pour la mise en place et la maintenance d'applications.

La mise en pratique des concepts présentés se fera en utilisant les fonctionnalités des SGBDs (PostgreSQL ou ORACLE) permettant de scruter et d'ajuster différents paramètres du système.

Le contenu du cours est le suivant :
1. Introduction
2. Gestion de la mémoire et stockage, index de hachage, index B,
3. Implémentation de SQL via les opérateurs relationnels,
4. Plan d'exécution.

Des notions élémentaires en bases de données relationnelles (conception de schémas, programmation SQL et fondement de SQL) sont nécessaires. Il est souhaitable mais non obligatoire d'avoir une connaissance de l'intégration de SQL avec un langage de programmation comme le C.

1. Système de gestion des bases de données, H. Korth et A. Silberschatz, McGraw-Hill. 2. Database Management Systems, Raghu Ramakrishnan et Johannes Gehrke, McGraw-Hill. 3. Database Systems: The Complete Book, by Hector Garcia-Molina, Jeffrey D. Ullman, and Jennifer Widom. Prentice Hall. 2002.

Septembre - Octobre.

ORSAY

Les travaux dirigés permettent de mettre en oeuvre les techniques étudiées en cours. Des lectures seront proposées aux étudiants pour compléter les interventions. Les travaux pratiques permettent d'expérimenter les différents outils mis à disposition par les systèmes pour scruter et agir sur les paramètres liés à la performance des applications.

L'objectif de ce cours est de dévoiler les mécanismes internes d'un SGBD permettant d'assurer la fiabilité et la qualité des données. L'accent est mis sur la gestion des transactions pour la concurrence d'accès et pour la reprise sur panne et d'aborder l'optimisation (tuning) des transactions pour les applications. Ces connaissances sont nécessaires autant aux développeurs d'application qu'aux administrateurs de base de données, par exemple pour la mise en place et la maintenance d'applications.

La mise en pratique des concepts présentés se fera en utilisant les fonctionnalités des SGBDs (PostgreSQL ou ORACLE) permettant de scruter et d'ajuster différents paramètres du système.

Le contenu du cours est le suivant :

1. Transactions
2. Mécanismes et protocoles de concurrence
3. Journalisation et reprise sur panne
4. Maintenance d'applications et tuning des transactions.

Des notions élémentaires en bases de données relationnelles (conception de schémas, programmation SQL et fondement de SQL) sont nécessaires. Il est souhaitable mais non obligatoire d'avoir une connaissance de l'intégration de SQL avec un langage de programmation comme le C.

1. Système de gestion des bases de données, H. Korth et A. Silberschatz, McGraw-Hill. 2. Database Management Systems, Raghu Ramakrishnan et Johannes Gehrke, McGraw-Hill. 3. Database Systems: The Complete Book, by Hector Garcia-Molina, Jeffrey D. Ullman, and Jennifer Widom. Prentice Hall. 2002.

Novembre - Décembre.

ORSAY

After a first introductory session, the students will present their progress, and their plans for future work. Objectives for the next session will be assigned accordingly. Some topics will be suggested by the teachers. Topics suggested by students may be accepted subject to approval by the teacher.

The goal of this project is to leverage the skills acquired in the other lectures in order to solve a data science problem, e.g., a recommendation system.
The student will combine multiple tools and data processing techniques to prepare, store, analyze and render data.
With respect to projects featured in other lectures, the project will emphasize this idea of combining multiple components within a long-term project.

Advanced databases, NoSQL databases, machine learning, data mining algorithms.

Novembre - Décembre.

ORSAY

De nombreux TPs permettront de se familiariser avec des systèmes représentatifs de différents types de traitements ou données, par exemple : batch processing avec Spark, fouille de documents JSON avec MongoDB, systèmes clé-valeurs avec Redis, gestion de flux de données avec Kafka ou bien un moteur de recherche textuel avec ElasticSearch.

Ce cours donne un panorama des différents systèmes de gestion de données distribués, et les concepts mis en œuvre dans ces systèmes.

Le cours abordera les points suivants : partitionnement, indexation, réplication, panorama des systèmes NoSQL, quelques éléments d'architecture et systèmes et structures de données et algorithmes utilisés dans ces systèmes.

Le cours abordera ces notions à travers de nombreux systèmes NoSQL, et en comparant ces systèmes avec les technologies relationnelles similaires.

Bases de données avancées.

Janvier - Février - Mars - Avril.

ORSAY

Le cours s'organise classiquement en séances de cours et de TD.

Ce cours est une introduction aux principes de formalisation et de de résolution de problèmes basés sur les méthodes de satisfaction (et/ou d'optimisation) de contraintes logiques, dont le champs d'application est très large. Il décrit les principes fondamentaux et les méthodes génériques que l'on retrouve au cœur des solveurs de problèmes de ce type, afin de pouvoir de familiariser avec leur utilisation.

Le cours introduit le concept de problème de satisfaction de contraintes et donne quelques principes pouvant guider la formalisation de tels problèmes en les illustrant sur des cas concrets. L'accent est mis sur le cas des problèmes à domaines finis. Il présente différentes notions de cohérence locale (cohérence de sommet, d'arc,...) ainsi que des algorithmes permettant de les établir, afin de pouvoir transformer automatiquement de tels problèmes en des problèmes plus simples. Il présente également les méthodes de recherche exhaustives permettant de trouver, énumérer les solutions de tels problèmes. L'intégration de méthodes de cohérence locale dans les méthodes de recherche exhaustive conduit à des méthodes de recherche hybrides dont les principes se retrouvent au cœur de la plupart des solveurs actuels.

Septembre - Octobre.

ORSAY

Le cours s'organise au format cours-TP, permettant d'illustrer rapidement les concepts introduits sur des exemples au fur et à mesure de leur introduction.

Ce cours permet de se familiariser avec l'utilisation pratique de solveurs génériques de contraintes. S'appuyant sur un solveur particulier, dont le langage est présenté en détail, il illustre son utilisation pour modéliser et résoudre des problèmes de complexité croissante. L'objectif est d'acquérir suffisamment d'autonomie pour pouvoir aborder la résolution d'un projet plus complexe.

Le cours permet illustrer de façon pratique, l'intérêt d'utiliser des solveurs génériques de problèmes de satisfaction/optimisation de contraintes. Il présente les caractéristiques (langage, outils de contrôle) d'un solveur de contraintes particulier tout en les illustrant sur différents problèmes concrets. Le cours est aussi l'occasion d'aborder quelques notions complémentaires liées à la résolution de tels problèmes, comme le paramétrage d'heuristiques, l'utilisation de contraintes globales et la modélisation de préférences.

Novembre - Décembre.

ORSAY

He course is taught in face to face lessons. The course is organised into three kinds of lessons: courses, exercises and technical sessions.

Today, we are experiencing an unprecedented production of resources, published as Linked Open Data (LOD, for short). This is leading to the creation of knowledge graphs (KGs) containing billions of RDF (Resource Description Framework) triples, such as DBpedia, YAGO and Wikidata on the academic side, and the Google Knowledge Graph or eBay Knowledge Graph on the commercial side. They contain knowledge that is typically expressed in RDF, i.e., as statements of the form . Sometimes, the various types and relations are represented in an OWL2 (Web Ontology Language) ontology, which defines their interrelations and axioms such as, subsumption, disjunction and functionality of properties.

However, the existing KGs are far from being complete and consistant. Hence, different methods are needed to be developed on top of these existing KGs. In one hand, methods that aim to expanding and enriching KGs, in the other hand, methods addressing the problem of validating the content of the KGs. In this course we will focus on the identity problem which consists in finding and validating identity links between resources and knowledge discovery problem (e.g. key axioms, logical rules) from RDF data. This course will also reserve place to some feedback from applications using knowledge graphs and ontologies such as bio-informatics, agronomy and IoT.

Septembre - Octobre - Novembre - Décembre.

ORSAY

Lectures, practical labs, and homework projects.

Searching information over rich web resources becomes a necessity for a large number of advanced applications. However, there are several impediments to use traditional keyword based search in practice due to the semantic mismatch among different resources. The course will introduce an approach to handle this problem, so called Semantic Web technology, and then it will focus on the knowledge representation aspect of Semantic Web, from W3C Semantic Web standards such as RDF, SPARQL and OWL to various representation formalisms (Description Logics) and their reasoning mechanisms.

Programming (Java), Propositional Logic.

Septembre - Octobre - Novembre - Décembre.

ORSAY

The course is organized in lecture classes and practical labs. The practical application of the concepts taught in the lectures will be applied during practical labs and assigned projects.

The course will teach students the basics of social and graph data management, and is organized in two parts. The first part will study graph metrics (degree distributions, clustering coefficients, distance metrics, etc.) with an objective to apply them to the analysis of real-graph data, especially social network graphs -- as found in Web application such as Facebook or Twitter -- and establish what makes them special compared to standard, random, graphs.

The second part of the course will focus on graph algorithms, as used for graph data analysis (PageRank, probabilistic reachability analysis), and apply them to a variety of applications such as link analysis, influence maximization, or link prediction. This part of the course is focused more on the practical aspect, and will be augmented by practical applications where the concepts will be applied. The course will present well-known systems for graph databases, such as Neo4J.
- applications to advertising and recommendation on the Web, such as collaborative filtering
- data stream mining: applying data mining algorithms for cases where the data is presented at a fast pace, and memory is insufficient to store (or access) the entirety of the dataset.

Advanced databases, Algorithms, Programming (Python/C/Java).

1. A.-L. Barabási. 'Network Science.' Cambridge University Press 2. M. Newman. 'Networks: An Introduction.' Oxford University Press 3. D. Easley, J. Kleingber. 'Networks, Crowds, and Markets.' Cambridge University Press.

Septembre - Octobre - Novembre - Décembre.

ORSAY

This course follows the HCI Bootcamp (Design of Interactive Systems), with an emphasis on redesign. Students will work in groups of three or four to design, evaluate and redesign an interactive system, represented as a video prototype. The goal is to significantly improve the initial design, based on input from users and other members of the class. Students will learn a variety of more advanced techniques, including co-adaptive instruments, interactive thread, participatory design workshops, cultural and technology probes, branching scenarios, generative walkthroughs, structured observation, peer interviewing, and alternative ways to present and explore a design space. All in-class and homework exercises are required to complete the final project, so class attendance and participation is essential. Grades are based on in-class and homework exercises as well as presentation of a final video prototype of each project, presented to and evaluated by an external jury.

ORSAY

Evaluation of the module: Project report, presentation and demonstration - 100%.

This module aims to provide students with advanced knowledge on last trend and innovative interactive techniques in the eXtended Reality (XR) domain for students who want to be a project manager or a researcher in this field. This module includes a set of 7 lecture sessions covering all mandatory aspects of multi-sensorimotor interactions for a fully immersive experience in virtual reality or between real and virtual worlds. Transversal knowledge about perceptual and cognitive issues which appear in XR, such as immersion, realism, presence, cybersickness will be also addressed.
The students will have to conduct an advanced XR project. From an end-user oriented expression of interests (either from their own focus of interest or from a list of typical applications provided by the teachers), a team of 3 to 4 students will have to choose the adapted XR hardware and software, to perform a team implementation and to present their work. This project will have to combine at least 2 of the 7 topics presented during the lectures.
Expected delivery: Slides and report presenting their project, the justification of design choices, a self-criticism of their solution/implementation, and a demonstration in front of the class. The project implementation will use the XR facilities of a dedicated tutorial room, and/or possibly some of the advanced equipment of the DIGISCOPE network.

Fundamentals of eXtended Reality (or equivalent).

ORSAY

This class focuses on advanced user interface programming techniques. Students will gain an understanding of the architecture and algorithmic underpinnings of modern UI toolkits. They will gain a rich operational knowledge of how to create user interfaces, including how to extend standard widgets and move beyond WIMP to off-the-desktop contexts. At the end of the course, students will be able to define new widgets and create new interactions.

ORSAY

Second part of [HCI] Advanced Programming of Interactive Systems 1.

ORSAY

Project work in a multi-disciplinary project with user cooperation in all phases of the project, from a general described theme to a specific and finished result. The project will be reported in different media. The project will be integrated with the Business Development Lab. Students will focus on an application area, do a market research and market segmentation, list the competitors, analyse weak and strong points, propose a new product concept, which will then be developed within the design project. After the design project, the prototype will be put to user testing, and the students will develop a business plan for marketing the product. During the BDL project, workshops and trainings will be provided by staff from the local centre of entrepreneurship / business school.

ORSAY

Second part of [HCI] Career Seminar - Level 1.

ORSAY

Throughout the course, the students will be able to: - Identify the key design principles across a range of related texts. - Participate substantively in discussions across the range of texts included in the course readings. - Apply these key principles in the design of their interactive artifact. - Conduct an in-depth project about the design of an interactive artwork. - Prepare and deliver a written paper demonstrating their application of the theoretical concepts in the design of the interactive artifact. - Present their interactive artifact during an exhibition.

This course requires students to read, reflect on a selection of authors' ideas, and then share their findings with the class. Weekly preparation for class includes a careful reading of all the assigned texts and short reading reports of maximum 200 words for each paper. The weekly readings will collectively provide a conceptual toolkit for the design of creative interactive art works. When reading an article, students need to look for the 'big picture' and important concepts that will inform their own work on the design of interactive artwork. Their task in this class is to identify and reflect on these high level concepts. They will be useful as both scholarly citations and as functional design tools for interactive works of art.
Students then need to design an interactive artifact that instantiate one or two theoretical concepts being discussed in the class. The artwork must be digital and interactive. The outcome of the classe will be shared during an afternoon exhibition of students work.

ORSAY

Projet part of [HCI] Creative Design.

ORSAY

Students will learn a principled approach to rapid prototyping of interactive systems, with an emphasis on iterative, cross-disciplinary design techniques. Lectures discuss the scientific and design-based foundations of participatory design. Students will work in small groups, applying these principles in a series of hands-on exercises that involve finding out about users, generating novel ideas, creating rapid prototypes and evaluating multiple design alternatives. Specific techniques include: critical incident interviews, video brainstorming, design spaces, paper and video prototyping, interaction tables, simple experiments, and design walkthroughs. The emphasis is on learning rapid prototyping techniques that produce design artifacts, which in turn support further design iterations. All in-class and homework exercises are required to complete the final project, so class attendance and participation is essential. Grades are based on in-class and homework exercises as well as presentation of a final video prototype of each project, presented to and evaluated by an external jury.

ORSAY

Project work in a multi-disciplinary project with user cooperation in all phases of the project, from a general described theme to a specific and finished result. The project will be reported in different media. The project will be integrated with the Business Development Lab. Students will focus on an application area, do a market research and market segmentation, list the competitors, analyse weak and strong points, propose a new product concept, which will then be developed within the design project. After the design project, the prototype will be put to user testing, and the students will develop a business plan for marketing the product. During the BDL project, workshops and trainings will be provided by staff from the local centre of entrepreneurship / business school.

ORSAY

Part 2 of the UE [HCI] Design project - Level 1.

ORSAY

Level 2 of the Design Project (see the description of [HCI] Design project - Level 1).

ORSAY

Second part of the UE [HCI] Design project - Level 2.

ORSAY

The Digital Fabrication Course is composed of a group of hands-on sessions and experiments at Fablab Digiscope, a digital fabrication facility dedicated to the prototyping of tangible objects and simple interactive devices at Université Paris-Saclay. The course will be composed of 3D Modeling and 3D printing sessions, 2D modeling and laser-cutting sessions and Electronics sessions with Arduino and sensors. This course offers students the possibility to develop their creativity and to increase their skills through a personal digital fabrication project related to their fields of interest. Students will also dive into the open-source and open-hardware movement by learning documentation technics to make their experience at Fablab Digiscope sharable and reproducible in any other fablab in the world.

The Digital Fabrication course is a series of experiments mobilizing the processes developed by the global community of +1500 fablabs. Contemporary approaches like Parametric Design, Short Iterations Design and Spiral Design will be collectively investigated to get an understanding of the Digital-Fabrication-Oriented Design (design for fabrication). Students will get the chance to access state of the art machinery like 3D Printers, laser-cutters, Vinyl-cutters and a fully equipped electronics bench...

ORSAY

Project part of [HCI] Digital fabrication.

ORSAY

This class is an introduction to the different methods for evaluating interactive systems. By the end of this class, students should be able to identify the right evaluation method to consider depending on the type of users, the type of the system and design stage of the system.
In particular, the following points will be discussed:
- Defining usability: measures, types of applications and types of users
- Capturing usability problems: questionnaires, heuristics and observational methods
- Operational tasks and predictive cognitive modeling
- Human factors and usability: basics of human anatomy, motor control and visual perception
- Evaluation "at large": longitudinal logging and crowdsourcing
- Introduction to experiment design
- Panorama of other evaluation methods in HCI literature: case studies, benchmarks

Expected Outcomes: Fundamental knowledge of available evaluation methods and challenges.

None.

ORSAY

This course focuses on laboratory experiments for testing an interactive system. Students will learn to define an experimental protocol: how to turn the phenomenon to study into a series of experiment tasks (operationalization), how to control potential biases in order to collect valid data. They will also learn good practices for setting up experiment environment and having people participate in their experiment. The course also details how to collect data and analyze them to draw valid conclusions. It both introduces 1) theory about descriptive and inferential statistics and 2) practical tools for running statistical tests with R Studio and reporting analyses.

ORSAY

Human-Computer Interaction addresses the design, development and evaluation of interactive systems. This area is quickly expanding as the number and diversity of users, devices, on-line services and available information increases. For example, up to 80% of the development cost of creating an interactive application is typically devoted to the user interface alone. This course introduces students to the foundations of Human-Computer Interaction (HCI), emphasizing its multi-disciplinary aspect. It successively covers a brief history of HCI, human factors and psychology of HCI, the design process of interactive systems, a review of interaction styles, with an emphasis on graphical interaction, and an overview of models and theories for HCI.

ORSAY

Second part of [HCI] Fundamental of Human-Computer Interaction 1.

ORSAY

Human-Computer Interaction addresses the design, development and evaluation of interactive systems. This research area is quickly expanding as the number and diversity of users, devices, on-line services and available information increases. For example, up to 80% of the development cost of creating an interactive application is typically devoted to the user interface alone.
This course introduces students to the foundations of Human-Computer Interaction (HCI), emphasizing its multidisciplinary aspect. It successively covers human factors and psychology of HCI, the design process of interactive systems, interaction styles and models, and evaluation methods, especially those based on controlled experiments.

ORSAY

7 * 3h (21h) for students in 2 groups Each 3h is divided in half-time for Lectures, half-time for Tutorial in VR/AR specific room. (for teacher that means 21h of Lectures, and 21h of Tutorials)

3 Lectures (3h each): - History of XR: Concepts, Taxonomies of devices and applications - 3D HCI for XR: Perceive, Think and Act in immersive environments - Programming for XR: Software, Principles and Constraints of XR programming 6 Tutorials (3h each): Design and Implementation of an XR application (project) Based on Unity software framework, teams of 2 to 3 students will choose within a number of projects proposals to implement an XR application based on available devices (such as HTC Vive, Microsoft HoloLens, Leapmotion, Haptic devices ...). Each project will be tested during the last tutorial session by other teams.

Total duration 21h for students in the 2 groups.

This module aims to give students all the fundamental knowledge necessary to understand the current state of Virtual, Mixed and Augmented Reality (VR, MR and AR), also called eXtended Reality (XR), and to master the basic software tools needed to design and develop applications in this field. Lectures are combined with hands-on tutorials to understand the main concepts of 3D immersive interaction while developing applied projects using the most recent technologies and software in XR. Tutorials will take place in a dedicated room equipped with XR facilities, in addition to the visits & demos in an advanced multi-sensorimotor and multi-user immersive environments (www.limsi.fr/venise/EVEsystem).

Expected Outcomes: Fundamental knowledge and background of XR interaction, with a perspective to work as a HCI engineer in that field.

None.

ORSAY

Gestural interaction is used as input modality in interfaces through a wide variety of gestures and movements: from deterministic and specific ones such as finger tapping on a touch screen to expressive whole-body interaction techniques. This course presents some techniques for capture, recognition and interpretation of gestures. Despite their diversity all these techniques have as common ground the human body with its capabilities and limitation for which interfaces must adapt and provide means of appropriate interaction.
The following topics will be covered in this course:
• Motor Control: biomechanics, movement control of the arm, hand and fingers, Fitts' law, etc.
• 2D interaction: touch-sensitive and multi-touch techniques, paper/pencil interaction, gestures discovery and memorization, Marking menus and derivatives, etc.
• 3D interaction: degrees of freedom, movement types, free hand gestures, gesture delimiters, embodied interaction, full-body interaction, etc.
• Gesture analysis: techniques for analyzing and recognizing gestures (automatons, supervised and unsupervised learning, HMMs, etc.)
• Applications to mobile interaction.

ORSAY

This course presents computer-supported collaborative systems, which allow a group of people, whether they are collocated or not, to work together while sharing computer artifacts. The course covers groupware and mediated interaction, including a state-of-the-art of interactive systems for coordination, communication and collaboration with groups of users across time and space. The course also covers Collaborative Virtual Environments, a research area at the intersection of Virtual Reality, Augmented Reality, teleoperation, high-bandwidth communication, human-computer interaction and collaborative teleworking. Finally it covers recent developments such as social networks and crowdsourcing.

ORSAY

Project part of [HCI] Groupware and Collaborative Work.

ORSAY

N recent years, we have seen information visualization tools receive more general adoption and integration in media, business, science, journalism, and many other fields. Information visualizations are used as essential tools for information analysis, exploration, or storytelling. This is in part the case because visual displays of information have several benefits. Looking at visual encodings of data has been shown to reduce search time, enhance detection of anticipated or unanticipated patterns, enable perceptual inference operations and hypothesis formulation, help the monitoring of changing data, and help data exploration by providing a manipulable medium. This course gives an overview of the field of information visualization. It involves a mixture of lectures, design exercises, programming assignments, and paper presentations. Students are expected to participate in class discussions and complete their assignments on time.

ORSAY

Projet part of [HCI] Interactive Information Visualization.

ORSAY

Machine learning based technology is now ubiquitous to many services and softwares that are used by humans on a daily basis. The common view of such technology is a black box that requires data only. However, humans developers and users contribute significantly in their design. A growing community at the intersection of Machine Learning and Human Computer Interaction seeks for interactive solutions in machine learning with the goal of improving the system performance, reducing the biases inherent to any machine learning systems, or offering explanations. The course starts with the basics of machine learning. Then we will focus on different approaches of putting human in the loop. We will review the seminal work in the field and do hands-on sessions via a software library developed for the class. The course is evaluated on project.

ORSAY

Projet part of [HCI] Interactive Machine Learning.

ORSAY

In this class we discuss "beyond-the-desktop" interaction, where visualization and interaction take place in the environment of the user, rather than through the traditional screen, mouse and keyboard. In particular, we cover the topic of mixed reality and augmented reality systems, which visually mix real and virtual objects; and that of tangible interaction, which moves interaction with online information to the objects around us and to the physical environment. We present a combination of techniques and devices to help augment the environment, such as spatial tracking and 3D visualization, mobile phones, glasses, hand-held projectors, etc. And a number of approaches to interacting with the environment that exploit the characteristics of physical objects or physical space, such as work on tangible objects, augmented paper, ambient displays, on-body interaction, interaction with large surfaces such as walls and tables, ubiquitous computing, etc.

ORSAY

Project part of [HCI] Mixed Reality and Tangible Interaction.

ORSAY

The course is an introduction to principles, methods and techniques relevant to the programming of interactive systems. This practical class will discuss available tools such as UI toolkits, and approaches for UI organization. Students will learn how to handle user input using event-based programming, as well as more advanced approaches such as state machines. During lab-hours, and as homework, students will work on programming exercises that apply the concepts and methods learned throughout the course.

Expected outcomes : Fundamental knowledge on UI programming.

Some knowledge of computer programming (javascript, python, C/C++, Java).

ORSAY

Second part of [HCI] Programming of Interactive Systems 1.

This is an interdisciplinary creative education program, in collaboration between Paris Saclay University, the diagonal Saclay and the Center Pompidou.

The course is conceived with the idea of ??becoming an annual meeting with artists and works from the collection of the National Museum of Modern Art at the Centre Pompidou as well as an opportunity to initiate projects between students and scientific disciplines. Students benefit each time from a theoretical and practical support to carry out the realization, production and diffusion of their project. Students will have the opportunity to share their experience during an interdisciplinary research meeting 'to be done" at the JACES and the Center Pompidou.".

It is a course on the design and implementation of experimental arts and sciences projects, led by a trio composed of two scientists and an artist.
Participants in this course will benefit from a practical and theoretical framework for the realization of projects combining science and / or technology and art. They will discover the collections of the Pompidou Center, guided by an artist. They will discover design methods based on experience and will have the opportunity on the one hand to develop their perceptual, cognitive and aesthetic sensitivity and on the other hand to deepen advanced technical and technological concepts, such as augmented reality, virtual reality, interactive and immersive spaces, sensors etc. They will be able to develop their imagination to design a work that integrates both a scientific and an aesthetic approach from their own personal stories.
The supervisors will guide the participants in formulating the intentions of their project in the form of short documentations that describe, analyze and discuss their conceptual, technical and critical choices. They will also accompany the resulting experiences of the participants and the spectators. Thus, the participants will have the responsibility to conceive a work based on an original concept and integrate a current scientific approach.

PARIS

Social interactions are becoming increasingly important in Human-Computer Interactions. Many systems involve virtual characters endowed with social interaction capabilities (video games, virtual worlds, animation, digital arts, pedagogical agents, on-line assistants, etc.). This module presents research in modeling, animating, and rendering realistic and expressive autonomous virtual humans. It introduces the theoretical foundations of multimodal communication in relation with theories and models from Psychology that are adapted to inform the design of computational models of interactive and expressive virtual humans. This includes theories, models and software architectures for managing emotions, personality, facial expressions, gestures and bodily interactions. Design frameworks and integration of these social dimensions in HCI design are presented. Applications in relevant fields of multimodal and social interaction are used as use cases: virtual coaches and persuasive technologies for sport, physical activities, nutrition, stress management, personal and corporate well being ; pedagogical agents for training social skills. Homework includes: reading scientific articles, watching a list of relevant TED like videos about fundamentals and applications of social interaction and virtual humans design, designing a project use case and step-by-step writing a scientific article summarizing the project.

ORSAY

Project part of [HCI] Virtual Humans.

ORSAY

L'objectif de ce cours est de fournir des outils et techniques algorithmiques de pointe aux apprentis. Etude de l'algorithmique sur les graphes (plus court chemin, tri topologique, …), les techniques de mémoisation, de programmation dynamique et de backtraching. Présentation de la notion de flots et des algorithmes de calcul de flot maximal. Enfin, les thèmes des algorithmes online et approchés seront abordés.

Janvier - Février - Mars - Avril.

ORSAY

Thomas Nowak, Université Paris Saclay.

Dans ce cours, il s'agira de concevoir et analyser des algorithmes distribués. Ce cours se basera sur les concepts de théorie des graphes étudiés en M1. Nous étudierons également la preuve de correction de ces algorithmes. Nous verrons comment comprendre les problématiques des systèmes distribuées, les différents algorithmes (élection, consensus, diffusion, …). Les solutions à ces problèmes seront étudiées ainsi qu'une application à différents problèmes.

Septembre - Octobre - Novembre - Décembre.

ORSAY

Le cours a pour but d'amener les apprentis à améliorer leurs compétences que ce soit à l'écrit et à l'oral. Les contenus abordés portent sur le monde du travail en général et le vocabulaire spécifique au domaine des systèmes d'information. Une préparation au TOEIC est également prévue.

Septembre - Octobre - Novembre - Décembre.

ORSAY

Le cours a pour but d'amener les apprentis à améliorer leurs compétences que ce soit à l'écrit et à l'oral. Les contenus abordés portent sur le monde du travail en général et le vocabulaire spécifique au domaine des systèmes d'information. Une préparation au TOEIC est également prévue.

Septembre - Octobre - Novembre - Décembre.

ORSAY

Ce cours combine des notions sur les marchés financiers (monnaies numériques, services bancaire, systèmes monétaires) et les concepts fondamentaux du développement des blockchains : architectures des systèmes distribués, consensus distribués, cryptographie, smart contracts.

Mars - Avril - Mai - Juin.

ORSAY

L'objectif de ce cours et d'apprendre à créer et déployer des applications Cloud. Le cours couvre les aspects suivants : Définition et types de cloud (cloud privé, public, hybride), modèles de services du cloud (SaaS, PaaS, LaaS), risques et opportunités du cloud (gestion de la sécurité, accès instantanés à des ressources de calcul et de stockage (élasticité), création et déploiement d'applications (Google Cloud, Amazon, Azure, ….).

Février - Mars - Avril - Mai - Juin.

ORSAY

L'objectif de ce cours est de donner d'une part les bases de la recherche scientifique : processus et méthodologie de la recherche scientifique, méthode de recherche bibliographique et sources bibliographiques scientifiques, principes de standardisation, étude et analyse de documents scientifiques et technologiques et d'autre part de présenter les principes de la communication et de l'information auprès des différents participants concernés par les projets menés.

Septembre - Octobre - Novembre - Décembre.

ORSAY

Ce module a pour objectif de présenter les nouveaux outils du domaine du Big Data. Il permettra de comprendre les solutions de stockage de données volumineuses et d'en faire des analyses. Il abordera quelques aspects préliminaires de la programmation distribuée ou Hadoop permettra de manipuler des données volumineuses pour aider à la prise de décision. Les étudiants verront également comment interroger de grandes masses de données en temps réel.

Mars - Avril - Mai - Juin.

ORSAY

Ce module couvre les technologies des BD relationnelles permettant de transformer des données issues de sources diverses en informations utiles à la prise de décision. Nous aborderons en particulier:
- modélisation:
au niveau conceptuel (modèles multidimensionnels) et niveau logique (schémas en étoiles...)
- le langage SQL et les optimisations de bases de données pour les requêtes OLAP:
indexes, partitionnement et vues
- l'architecture des entrepôts
- les processus ETL (rapidement)
- les grands principes des bases de données 'in memory'
Pré-requis (couvert par le programme de Licence) :
- programmation en Java
- BD relationnelles : schéma et notions associées (clé primaires, clé
étrangères, contraintes, …), requêtes en SQL et mises à jour.

Septembre - Octobre - Novembre - Décembre.

ORSAY

Ce cours est la suite du cours Data Warehouse I. Il met plus l'accent sur les ETL.

Mars - Avril - Mai - Juin.

ORSAY

Cours sur le droit informatique, la rgpd.

Mars - Avril - Mai - Juin.

ORSAY

De nombreuses applications sont confrontées avec une grande quantité d'informations numériques qui nécessitent un traitement statistique. Cet enseignement présente les principales techniques d'analyse et d'interprétation de données numériques telles que la séparation linéaire, la classification Bayesienne, le regroupement automatique et les réseaux connexionnistes. La deuxième partie de cet enseignement présente des algorithmes qui sont utilisés en traitement du langage parlé et en reconnaissance de geste ou d'écriture. Après une introduction à la technique de la programmation dynamique, l'utilisation et l'apprentissage des modèles de Markov cachés seront traités en détail.

Février - Mars - Avril - Mai.

ORSAY

Ce module a pour objectif de familiariser les étudiants avec des approches basiques de machine learning. Le cours commencera en présentant les modèles traditionnels d'apprentissage supervisé et non-supervisé (tel que l'algorithme des k-moyennes, la décomposition en valeurs principales, le percetron, ...) pour entrevoir par la suite des modèles plus complexes. Le cours presentera les aspects théoriques du machine learning, et une grand part de travaux pratiques sera mis à disposition pour appliquer les techniques vues en cours à divers problèmes (classification, partitionnement, filtrage collaboratif, ...).

Octobre - Novembre - Décembre.

ORSAY

L'Internet des objets représentent l'extension d'Internet à des dispositifs et à des lieux du monde physique. Il est considéré comme la troisième évolution de l'Internet, baptisée Web 3.0 qui fait suite à l'ère du web social. On parle d'objets connectés pour définir des types d'objets auxquels l'ajout d'une connexion Internet a permis d'apporter une valeur supplémentaire en termes de fonctionnalité, d'information, d'interaction avec l'environnement ou d'usage. Ce module présente le concept d'objets et de l'IoT, les fonctionnalités et technologies de communication, la mise en réseau et les architectures dédiées, les domaines d'application et nouveaux services, la standardisation et le lien entre l'IoT et le le Big Data.

Février - Mars - Avril - Mai.

ORSAY

Dans ce cours, l'accent sera mis sur les langages dits dynamiques, c'est à dire des langages qui exécutent des actions au moment de l'exécution alors que d'autres langages les exécutent à la compilation. Des langages tels que Javascript, Python, PHP, Ruby seront étudiés. Des travaux pratiques permettront de mettre en œuvre les concepts liés à ces langages qui seront vus pendant le cours.

Septembre - Octobre - Novembre - Décembre.

ORSAY

L'objectif de ce cours est de donner les bases indispensables en apprentissage automatique ou "Machine Learning": les principales familles de modèles et les algorithmes associés (inférence et apprentissage). En particulier, seront traités : (i) les bases théoriques de l'apprentissage, les modèles probabilistes et l'apprentissage Bayésien, (ii) les modèles linéaires et les SVM (Support Vector Machines).

Octobre - Novembre - Décembre - Janvier.

ORSAY

A partir de son expérience, des enseignements reçus, de la mission à mener à bien en entreprise, l'apprenti effectue un travail de recherche personnel aboutissant à un mémoire qui devra être rédigé et présenté devant un jury en fin de cursus.

Janvier - Février - Mars - Avril - Mai - Juin - Juillet.

ORSAY

L'objectif de ce cours est de fournir un tour d'horizon des modèles mathématiques possibles. Ecriture sous formes mathématiques de problèmes réels prenant en compte le temps (modèles avec équations différentielles), des données certaines (programmes mathématiques linéaires et non linéaires en variables mixtes), des données incertaines (programmation robuste et stochastique), plusieurs décideurs (modèles à choix discrets).

Septembre - Octobre - Novembre - Décembre.

ORSAY

Dans ce cours, sont abordés les langages de modélisation tels que SDL, BPMN. Il s'agit de sensibiliser les étudiants à la nécessité de réfléchir à un problème posé avant de le programmer. Ces langages permettent de décrire des systèmes, des processus métier. BPMN permet de modéliser des processus non informatiques. Des choix devront être faits pour se concentrer sur l'analyse de ces systèmes. Ces choix porteront sur le niveau d'abstraction, les choix des interactions, des algorithmes. Les détails liés à l'implémentation seront laissés de côté pour la phase de conception.

Septembre - Octobre - Novembre - Décembre.

ORSAY

Le cours porte sur les sujets suivants : Bornes, optimalité et relaxation; problèmes d'affectation et de couplages, programmation dynamique, Branch and Bound.

Septembre - Octobre - Novembre - Décembre.

ORSAY

Les données échangées sont souvent exploitées par les différents systèmes en fonction de représentation interne de chaque logiciel. Il est dont souvent nécessaire de produire des analyseurs syntaxiques et lexicaux pour manipuler ces données. Nous verrons dans ce cours comment construire des analyseurs en utilisant Lex et Yacc. Nous verrons également comment manipuler des textes avec sed et awk. Enfin, nous verrons également comment automatiser sous forme de script batch ou autres des tâches.

Septembre - Octobre - Novembre - Décembre.

ORSAY