Program Year
UE obligatoires
-
Fondamentaux de l’apprentissage automatique
Fondamentaux de l’apprentissage automatique
Ects : 6
Lecturer :
YANN CHEVALEYRE
Total hours : 48
Overview :
The course will introduce the theoretical foundations of machine learning, review the most successful algorithms with their theoretical guarantees, and discuss their application in real world problems. The covered topics are:
- Part 1: Supervised Learning Theory: the batch setting IntroSurrogate LossesUniform Convergence and PAC Learning Empirical Risk Minimization and ill-posed problemsConcentration InequalitiesUniversal consistency, PAC LearnabilityVC dimensionRademacher complexityNon Uniform Learning and Model Selection biais-variance tradeoffStructural Minimization Principle and Minimum Description Length PrincipleRegularization
- Part 2: Supervised Learning Theory and Algorithms in the Online Setting Foundations of Online LearningBeyond the Perceptron algorithm
- Partie 3: Ensemble Methods and Kernels Methods SVMs, KernelsKernel approximation algorithms in the primalEnsemble methods: bagging, boosting, gradient boosting, random forests
- Partie 4: Algorithms for Unsupervised Learning Dimensionality reduction: PCA, ICA, Kernel PCA, ISOMAP, LLERepresentation LearningExpectation Maximization, Latent models and Variational methods
Recommended prerequisites :
- Linear models
Require prerequisites :
- Linear Algebra - Statistics and Probability
Learning outcomes :
The aim of this course is to provide the students with the fundamental concepts and tools for developing and analyzing machine learning algorithms.
Assessment :
- Each student will have to have the role of scribe during one lecture, taking notes during the class and sending the notes to the teacher in pdf. - Final exam
Bibliography-recommended reading
The most important book: - Shalev-Shwartz, S., & Ben-David, S. (2014). Understanding machine learning: From theory to algorithms. Cambridge university press. Also: - Mohri, M., Rostamizadeh, A., & Talwalkar, A. (2012). Foundations of machine learning. MIT press. - Vapnik, V. (2013). The nature of statistical learning theory. Springer science & business media. - Bishop Ch. (2006). Pattern recognition and machine learning. Springer - Friedman, J., Hastie, T., & Tibshirani, R. (2001). The elements of statistical learning (Vol. 1, No. 10). New York, NY, USA:: Springer series in statistics. - James, G., Witten, D., Hastie, T., & Tibshirani, R. (2013). An introduction to statistical learning (Vol. 112). New York: springer.
-
Optimization for Machine Learning
Optimization for Machine Learning
Ects : 6
Lecturer :
GABRIEL PEYRE
Total hours : 48
Overview :
This course will review the mathematical foundations for Machine Learning, as well as the underlying algorithmic methods and showcases some modern applications of a broad range of optimization techniques.
Optimization is at the heart of most recent advances in machine learning. This includes of course most basic methods (linear regression, SVM and kernel methods). It is also the key for the recent explosion of deep learning which are state of the art approaches to solve supervised and unsupervised problems in imaging, vision and natural language processing. This course will review the mathematical foundations, the underlying algorithmic methods and showcases some modern applications of a broad range of optimization techniques.
The course will be composed of both classical lectures and numerical sessions in Python. The first part covers the basic methods of smooth optimization (gradient descent) and convex optimization (optimality condition, constrained optimization, duality). The second part will features more advanced methods (non-smooth optimization, SDP programming,interior points and proximal methods). The last part will cover large scale methods (stochastic gradient descent), automatic differentiation (using modern python framework) and their application to neural network (shallow and deep nets).
Bibliography-recommended reading
Theory and algorithms: Convex Optimization, Boyd and Vandenberghe Introduction to matrix numerical analysis and optimization, Philippe Ciarlet Proximal algorithms, N. Parikh and S. Boyd Introduction to Nonlinear Optimization - Theory, Algorithms and Applications, Amir Beck Numerics: Pyrthon and Jupyter installation: use only Python 3 with Anaconda distribution. The Numerical Tours of Signal Processing, Gabriel Peyré Scikitlearn tutorial #1 and Scikitlearn tutorial #2, Fabian Pedregosa, Jake VanderPlas Reverse-mode automatic differentiation: a tutorial Convolutional Neural Networks for Visual Recognition Christopher Olah, Blog
-
Bases de données avancées (SBGD non classiques)
Bases de données avancées (SBGD non classiques)
Ects : 6
Lecturer :
KHALID BELHAJJAME
DANIELA GRIGORI
DARIO COLAZZO
Total hours : 48
Overview :
It is well known that in the context of data processing for IA applications, a large part of the effort is devoted to data preparation, a process that strongly depends on techniques and skills for formulating complex queries and tuning systems supporting their execution, in order to ensure reasonable query execution time. In a wide range of use cases, data preparation involves either structured or semi-structured datasets.
In this context, the first goal of the course is to enable students to acquire strong skills for the efficient querying of relational databases. After an initial refresh about the standard query language and the design of complex queries, the course will present optimization techniques that relational database management systems adopt in order to ensure efficient querying. The attention will be particularly given to storage and indexing techniques, as well as algorithms for the generation of efficient query execution plans, and main approach for database tuning. How these techniques are transposed in modern systems of the big data ecosystem will be also discussed.
The second goal of the course is to present foundations and advanced techniques supporting systems for semi-structured data processing. The attention is first focused formal specification of query languages, the design of complex queries as well as recent techniques for static analysis that help the user in the design of correct queries, in a context where this task is particularly difficult due to the potentially high and unpredictable variability in the structure of the datasets. Both parts will be supported by both books and scientific articles published in main conferences and journal on databases. The acquired notions will be consolidated in several lab-sessions.
Learning outcomes :
The first goal of the course is to enable students to acquire strong skills for the efficient querying of relational databases. The second goal of the course is to present foundations and advanced techniques supporting systems for semi-structured data processing.
Bibliography-recommended reading
Database Management Systems, Third Edition, Raghu Ramakrishnan, Johannes Gehrke. Mac Graw Hill Relational DBMS internals. Antonio Albano, Dario Colazzo, Giorgio Ghelli, Enzo Orsini. Book available in pdf.
-
Représentation des connaissances, raisonnement, planification
Représentation des connaissances, raisonnement, planification
Ects : 4
Lecturer :
GABRIELLA PIGOZZI
JEROME LANG
Total hours : 24
Overview :
The course introduces techniques for representing and reasoning over knowledge information.
1. Reasoning about Belief, Knowledge, and Preferences
- plausible and nonmonotonic reasoning
- reasoning about belief and knowledge (single-and multiple-agent), belief change
- case-based reasoning, analogical reasoning
- preference languages, reasoning about preferences
- reasoning and decision under uncertainty, graphical models
2. Planning
- reasoning about action, action languages for planning
- algorithms for classical planning
- short introduction to decision theory and decision-theoretic planning
- planning under uncertainty and full observability: Markov decision processes
- planning under partial observabilty: partially observable Markov decision processes
- [time permitting] multi-agent planning
Recommended prerequisites :
none
Require prerequisites :
none.
Learning outcomes :
N/A
Assessment :
written exam.
-
Projet Sciences des Données
Projet Sciences des Données
Ects : 4
Lecturer :
BENJAMIN NEGREVERGNE
Total hours : 24
Overview :
Students enrolled in this class will form groups and choose one topic among a list of proposed topics in the core areas of the master such as supervised or unsupervised learning, recommendation, game AI, distributed or parallel data-science, etc. The topics will generally consist in applying a well-established technique on a novel data-science challenge or in applying recent research results on a classical data-science challenge. Either way, each topic will come with its own novel scientific challenge to address. At the end of the module, the students will give an oral presentation to demonstrate their methodology and their findings. Strong scientific rigor as well as very good engineering and communication skills will be necessary to complete this module successfully.
Learning outcomes :
The goal of this module is to provide students with a hands-on experience on a novel data-science/AI challenge using state-of-the-art tools and techniques discussed during other classes of this master.
-
Apprentissage Profond
Apprentissage Profond
Ects : 4
Lecturer :
TRISTAN CAZENAVE
Total hours : 24
Overview :
Introduction to Deep Learning. Deep Learning enable to train neural network with many layers so as to address various difficult problems. Applications range from image to games. In this course we will present Stochastic Gradient Descent for deep neural networks using different architectures (convolutions, dense, recurrent, residual). We will use Keras/Tensorflow and/or Pytorch and apply them to games and optimization. References: Deep Learning avec TensorFlow - Mise en oeuvre et cas concrets – 22 novembre 2017 de Aurélien Géron, O’Reilly. Keras Documentation : keras.io Pytorch Documentation : pytorch.org
UE optionnels
-
Apprentissage automatique avancé
Apprentissage automatique avancé
Ects : 3
Lecturer :
YANN CHEVALEYRE
Total hours : 24
Overview :
This research-oriented module will focus on advanced machine learning algorithms, in particular in the Bayesian setting
1) Bayesian Machine Learning (with Moez Draief, chief data scientist CapGemini) - Bayesian linear regression - Gaussian Processes (i.e. kernelized Bayesian linear regression) - Approximate Bayesian Inference - Latent Dirichlet Allocation 2) Bayesian Deep Learning (with Julyan Arbel, CR INRIA) - MCMC methods - variationnal methods 3) Advanced Recommandation Techniques (with Clement Calauzene, Criteo)
Learning outcomes :
Probabilistic, Bayesian ML and recommandation systems
Assessment :
- Chaque étudiant devra présenter un papiers de recherche
-
Apprentissage incrémental, Théorie des Jeux et Applications
Apprentissage incrémental, Théorie des Jeux et Applications
Ects : 3
Total hours : 24
Overview :
This course will focus on the behavior of learning algorithms when several agents are competing against one another: specifically, what happens when an agent that follows an online learning algorithm interacts with another agent doing the same? The natural language to frame such questions is that of game theory, and the course will begin with a short introduction to the topic, such as normal form games (in particular zero-sum, potential, and stable games), solution concepts (such as dominated/rationalizable strategies, Nash, correlated and coarse equilibrium notions, ESS), and some extensions (Blackwell approachability). Subsequently, we will examine the long-term behavior of a wide variety of online learning algorithms (fictitious play, regret-matching, multiplicative/exponential weights, mirror descent and its variants, etc.), and we will discuss applications to generative adversarial networks (GANs), traffic routing, prediction, and online auctions.
[1] Nicolò Cesa-Bianchi and Gábor Lugosi, Prediction, learning, and games, Cambridge University Press, 2006. [2] Drew Fudenberg and David K. Levine, The theory of learning in games, Economic learning and social evolution, vol. 2, MIT Press, Cambridge, MA, 1998. [3] Sergiu Hart and Andreu Mas-Colell, Simple adaptive strategies: from regret matching to uncoupled dynamics, World Scientific Series in Economic Theory - Volume 4, World Scientific Publishing, 2013. [4] Vianney Perchet, Approachability, regret and calibration: implications and equivalences, Journal of Dynamics and Games 1 (2014), no. 2, 181–254. [5] Shai Shalev-Shwartz, Online learning and online convex optimization, Foundations and Trends in Machine Learning 4 (2011), no. 2, 107–194.
-
Apprentissage profond pour l’analyse d’images
Apprentissage profond pour l’analyse d’images
Ects : 3
Lecturer :
Etienne DECENCIERE
Total hours : 24
Overview :
Deep learning has achieved formidable results in the image analysis field in recent years, in many cases exceeding human performance. This success opens paths for new applications, entrepreneurship and research, while making the field very competitive.
This course aims at providing the students with the theoretical and practical basis for understanding and using deep learning for image analysis applications.
Program to be followed The course will be composed of lectures and practical sessions. Moreover, experts from industry will present practical applications of deep learning. Lectures will include: •Artificial neural networks, back-propagation algorithm • Convolutional neural network • Design and optimization of a neural architecture • Successful architectures (AlexNet, VGG, GoogLeNet, ResNet) • Analysis of neural network function • Image classification and segmentation • Auto-encoders and generative networks • Current research trends and perspectives
During the practical sessions, the students will code in Python, using Keras and Tensorflow. They will be confronted with the practical problems linked to deep learning: architecture design; optimization schemes and hyper-parameter selection; analysis of results.
Require prerequisites :
- Linear algebra, basic probability and statistics
- Python
Learning outcomes :
Deep learning: theoretical foundations and applications
Assessment :
Exam
-
Traitement Automatique des Langues
Traitement Automatique des Langues
Ects : 3
Lecturer :
ALEXANDRE ALLAUZEN
Total hours : 24
Overview :
The course focuses on modern and statistical approaches to NLP.
Natural language processing (NLP) is today present in some many applications because people communicate most everything in language : post on social media, web search, advertisement, emails and SMS, customer service exchange, language translation, etc. While NLP heavily relies on machine learning approaches and the use of large corpora, the peculiarities and diversity of language data imply dedicated models to efficiently process linguistic information and the underlying computational properties of natural languages.
Moreover, NLP is a fast evolving domain, in which cutting-edge research can nowadays be introduced in large scale applications in a couple of years.
The course focuses on modern and statistical approaches to NLP: using large corpora, statistical models for acquisition, disambiguation, parsing, understanding and translation. An important part will be dedicated to deep-learning models for NLP.
- Introduction to NLP, the main tasks, issues and peculiarities - Sequence tagging: models and applications - Computational Semantics - Syntax and Parsing - Deep Learning for NLP: introduction and basics - Deep Learning for NLP: advanced architectures - Deep Learning for NLP: Machine translation, a case study
Bibliography-recommended reading
References - Costa-jussà, M. R., Allauzen, A., Barrault, L., Cho, K., & Schwenk, H. (2017). Introduction to the special issue on deep learning approaches for machine translation. Computer Speech & Language, 46, 367-373. - Dan Jurafsky and James H. Martin. Speech and Language Processing (3rd ed. draft): web.stanford.edu/~jurafsky/slp3/ - Yoav Goldberg. A Primer on Neural Network Models for Natural Language Processing: u.cs.biu.ac.il/~yogo/nnlp.pdf - Ian Goodfellow, Yoshua Bengio, and Aaron Courville. Deep Learning: www.deeplearningbook.org
-
Nuages de Points et Modélisation 3D
Nuages de Points et Modélisation 3D
Ects : 3
Lecturer :
Francois GOULETTE
Total hours : 24
Overview :
Ce cours donne un panorama des concepts et techniques d'acquisition, de traitement et de visualisation des nuages de points 3D, et de leurs fondements mathématiques et algorithmiques.
Le cours abord notamment les thème suivants : Systèmes de perception 3D Traitements et opérateurs Recalage Segmentation de nuages de points Reconstruction de courbes et surfaces Modélisation par primitives Rendu de nuages de points et maillages
La plupart des séances sont complétées d'un TP.
Les cours sont en français, les sujets des TP sont en anglais.
Site Web : caor-mines-paristech.fr/fr/cours-npm3d/
Learning outcomes :
Compétences théoriques et pratiques sur la production, le traitement et les applications des nuages de points 3D.
Assessment :
Comptes-rendus de TP.
Projet sur article.
-
Ethique et intelligence artificielle
Ethique et intelligence artificielle
Ects : 3
Lecturer :
Francois GOULETTE
Total hours : 24
Overview :
Objectifs pédagogiques
Sensibiliser un public essentiellement technique (informaticiens) : – aux transformations profondes en cours avec la révolution numérique – aux cadres juridiques et règlementaires déjà existants et comment les respecter – aux questions éthiques ouvertes, et à la réflexion éthique pour des questions à venir
Site Web : caor-mines-paristech.fr/fr/cours-eia/
Learning outcomes :
Compétences d'analyse des aspects éthiques, juridiques et techniques sur nouveaux produits et services utilisant l'Intelligence Artificielle.
Assessment :
Projets sur des cas concrets de produits ou services utilisant de l'IA (ou nouvelles technologies au sens large), à réaliser en équipe.
-
Graphes de connaissance, logiques de description, raisonnement sur les données
Graphes de connaissance, logiques de description, raisonnement sur les données
Ects : 3
Lecturer :
Michael THOMAZZO
Total hours : 24
Overview :
Introduction to Knowledge Graphs, Description Logics and Reasoning on Data.
Knowledge graphs are a flexible tool to represent knowledge about the real world. After presenting some of the existing knowledge graphs (such as DBPedia, Wikidata or Yago) , we focus on their interaction with semantics, which is formalized through the use of so-called ontologies. We then present some central logical formalism used to express ontologies, such as Description Logics and Existential Rules. A large part of the course will be devoted to study the associated reasoning tasks, with a particular focus on querying a knowledge graph through an ontology. Both theoretical aspects (such as the tradeoff between the expressivity of the ontology language versus the complexity of the reasoning tasks) and practical ones (efficient algorithms) will be considered.
Program: 1. Knowledge Graphs (history and uses) 2. Ontology Languages (Description Logics, Existential Rules) 3. Reasoning Tasks (Consistency, classification, Ontological Query Answering) 4. Ontological Query Answering (Forward and backward chaining, Decidability and complexity, Algorithms, Advanced Topics)
References: -- The description logic handbook: theory, implementation, and applications. Baader et al., Cambridge University Press -- Foundations of Semantic Web Technologies, Hitzler et al., Chapman&Hall/CRC -- Web Data Management, Abiteboul et al., Cambridge University Press
Prerequisites: -- first-order logic; -- complexity (Turing machines, classical complexity classes) is a plus.
-
Fouille de Graphes
Fouille de Graphes
Ects : 3
Lecturer :
DANIELA GRIGORI
Total hours : 24
Overview :
The objective of this course course is to give students an overview of the field of graph analytics. Since graphs form a complex and expressive data type, we need methods for representing graphs in databases, manipulating, querying, analyzing and mining them. Moreover, graph applications are very diverse and need specific algorithms. The course presents new ways to model, store, retrieve, mine and analyze graph-structured data and some examples of applications. Lab sessions are included allowing students to practice graph analytics: modeling a problem into a graph database and performing analytical tasks over the graph in a scalable manner.
Program
• Graph analytics
– Networks properties and models
– Link Analysis : PageRank and its variants
– Community detection
• Frameworks for parallel graph analytics
– Pregel – a model for parallel-graph computing
– GraphX Spark – unifying graph- and data –parallel computing
• Machine learning with graphs
• Applications : process mining and analysis
Practical work : graph analytics with GraphX and Neo4J
Learning outcomes :
Modeling a problem into a graph model and performing analytical tasks over the graph in a scalable manner.
Bibliography-recommended reading
References
Ian Robinson, Jim Weber, Emil Eifrem, Graph Databases, Editeur : O'Reilly (4 juin 2013), ISBN-10: 1449356265
Eric Redmond, Jim R. Wilson, Seven Databases in Seven Weeks - A Guide to Modern Databases and the NoSQL Movement, Publisher: Pragmatic Bookshelf
Grzegorz Malewicz, Matthew H. Austern, Aart J.C Bik, James C. Dehnert, Ilan Horn, Naty Leiser, and Grzegorz Czajkowski. 2010. Pregel: a system for large-scale graph processing, SIGMOD '10, ACM, New York, NY, USA, 135-146
Xin, Reynold & Crankshaw, Daniel & Dave, Ankur & Gonzalez, Joseph & J. Franklin, Michael & Stoica, Ion. (2014). GraphX: Unifying Data-Parallel and Graph-Parallel Analytics.
Michael S. Malak and Robin East, Spark GraphX in Action, Manning, June 2016
-
Machine Learning sur Big Data
Machine Learning sur Big Data
Ects : 3
Lecturer :
DARIO COLAZZO
Total hours : 24
Overview :
This course focuses on the typical, fundamental aspects that need to be dealt with in the design of machine learning algorithms that can be executed in a distributed fashion, typically on Hadoop clusters, in order to deal with big data sets, by taking into account scalability and robustness.
Nowadays there is an ever increasing demand of machine learning algorithms that scales over massives data sets. In this context, this course focuses on the typical, fundamental aspects that need to be dealt with in the design of machine learning algorithms that can be executed in a distributed fashion, typically on Hadoop clusters, in order to deal with big data sets, by taking into account scalability and robustness. So the course will first focus on a bunch of main-stream, sequential machine learning algorithms, by taking then into account the following crucial and complex aspects. The first one is the re-design of algorithms by relying on programming paradigms for distribution and parallelism based on map-reduce (e.g., Spark, Flink, ….). The second aspect is experimental analysis of the map-reduce based implementation of designed algorithms in order to test their scalability and precision. The third aspect concerns the study and application of optimisation techniques in order to overcome lack of scalability and to improve execution time of designed algorithm.
The attention will be on machine learning technique for dimension reduction, clustering and classification, whose underlying implementation techniques are transversal and find application in a wide range of several other machine learning algorithms. For some of the studied algorithms, the course will present techniques for a from-scratch map-reduce implementation, while for other algorithms packages like Spark ML will be used and end-to-end pipelines will be designed. In both cases algorithms will be analysed and optimised on real life data sets, by relaying on a local Hadoop cluster, as well as on a cluster on the Amazon WS cloud.
References:
- Mining of Massive Datasets www.mmds.org
- High Performance Spark - Best Practices for Scaling and Optimizing Apache Spark Holden Karau, Rachel Warren O'Reilly
-
Choix social computationnel
Choix social computationnel
Ects : 3
Lecturer :
JEROME LANG
Total hours : 24
Overview :
The aim of this course is to give an overview of the problems, techniques and applications of computational social choice, a multidisciplinary topic at the crossing point of computer science (especially artificial intelligence, operations research, theoretical computer science, multi-agent systems, computational logic, web science) and economics.
The course consists of the analysis of problems arising from the aggregation of preferences of a group of agents from a computational perspective. On the one hand, it is concerned with the application of techniques developed in computer science, such as complexity analysis or algorithm design, to the study of social choice mechanisms, such as voting procedures or fair division algorithms. On the other hand, computational social choice is concerned with importing concepts from social choice theory into computing.
The course will focus on normative aspects, computational aspects, and real-world applications (including some case studies).
Program:
1. Introduction to social choice and computational social choice.
2. Preference aggregation, Arrow's theorem and how to escape it.
3. Voting rules: informational basis and normative aspects.
4. Voting rules : computation.
5. Strategic issues: strategyproofness, Gibbard and Satterthwaite's theorem, computational resistance to manipulation, other forms of strategic behaviour.
6. Voting on combinatorial domains.
7. Communication issues in voting: voting with incomplete preferences, elicitation protocols, communication complexity, low-communication social choice.
8. Fair division of indivisible goods.
9. Matching under preferences.
10. Specific applications and case studies (varying every year): rent division, kidney exchange, school assignment, group recommendation systems…
Recommended prerequisites :
Prerequisite-free. Basics of discrete mathematics (especially graph theory) and algorithmics is a plus.
Require prerequisites :
none
Learning outcomes :
N/S
Assessment :
Written exam by default. Possibility to replace the exam by a project.
Bibliography-recommended reading
References: * Handbook of Computational Social Choice (F. Brandt, V. Conitzer, U. Endriss, J. Lang, A. Procaccia, eds.), Cambridge University Press, 2016. Available for free online. * Trends in Computational Social Choice (U. Endriss, ed), 2017. Available for free online.
-
Recherche Monte-Carlo et Jeux
Recherche Monte-Carlo et Jeux
Ects : 3
Lecturer :
TRISTAN CAZENAVE
Total hours : 24
Overview :
Introduction to Monte Carlo for computer games.
Monte Carlo Search has revolutionized computer games. It works well with Deep Learning so as to create systems that have superhuman performances in games such as Go, Chess, Hex or Shogi. It is also appropriate to address difficult optimization problems. In this course we will present different Monte Carlo search algorithms such as UCT, GRAVE, Nested Monte Carlo and Playout Policy Adaptation. We will also see how to combine Monte Carlo Search and Deep Learning. The validation of the course is a project involving a game or an optimization problem.
La recherche Monte-Carlo a révolutionné la programmation des jeux. Elle se combine bien avec le Deep Learning pour créer des systèmes qui jouent mieux que les meilleurs joueurs humains à des jeux comme le Go, les Echecs, le Hex ou le Shogi. Elle permet aussi d’approcher des problèmes d’optimisation difficiles. Dans ce cours nous traiterons des différents algorithmes de recherche Monte-Carlo comme UCT, GRAVE ou le Monte-Carlo imbriqué et l’apprentissage de politique de playouts. Nous verrons aussi comment combiner recherche Monte-Carlo et apprentissage profond. Le cours sera validé par un projet portant sur un jeu ou un problème d’optimisation difficile.
Bibliography-recommended reading
Bibliographie : Intelligence Artificielle Une Approche Ludique, Tristan Cazenave, Editions Ellipses, 2011.
-
Introduction to reinforcement learning
Introduction to reinforcement learning
Ects : 3
Total hours : 24
Overview :
1. Multiarmed bandits, Markov Decision Processes and other models 2. Planning: finite and infinite horizon problems, the value function, Bellman equations, dynamic programming, value and policy iteration 3. Probabilistic and statistical tools for RL: Bayesian models, relative entropy and hypothesis testing, concentration inequalities, linear regression, the stochastic approximation algorithm 4. RL algorithms for multiarmed bandits: the explore vs. exploit compromise, bandit algorithms vs. A/B testing, UCB, Thomson sampling, contextual bandits 5. RL algorithms for Markov Decision Processes: off policy and on policy learning, Q-learning, SARSA, Monte Carlo tree search
Learning outcomes :
This introductory course will provide the main methodological building blocks of reinforcement learning. Some basic notions in probability theory are required to follow the course. The course will imply some work on simple implementations of the algorithms, assuming familiarity with common scientific computing language.
Bibliography-recommended reading
Bibliographie, lectures recommandées M. Puterman. Markov Decision Processes: Discrete Stochastic Dynamic Programming. John Wiley & Sons, 1994. R. Sutton and A. Barto. Introduction to Reinforcement Learning. MIT Press, 1998. C. Szepesvari. Algorithms for Reinforcement Learning. Morgan & Claypool Publishers, 2010 J. Myles White. Bandit Algorithms for Website Optimization. O'Reilly. 2012 T. Lattimore and C. Szepesvari. Bandit Algorithms. Cambridge University Press. 2019. downloads.tor-lattimore.com/banditbook/book.pdf
-
Deep renforcement learning et applications
Deep renforcement learning et applications
Overview :
What you will learn in this class?
- Intro and Course Overview
- Supervised Learning behaviors
- Principles of Reinforcement Learning
- Policy Gradients
- Actor-Critic Algorithms (A2C, A3C and Soft AC)
- Value Function Methods
- Deep RL with Q-functions
- Advanced Policy Gradient (DDPG, Twin Delayed DDPG)
- Trust Region & Proximal Policy Optimization (TRPO, PPO)
- Optimal Control and Planning
- Model-Based Reinforcement Learning
- Model-Based Policy Learning
- Exploration and Stochastic Bandit in RL
- Exploration with Curiosity and Imagination
- Offline RL and Generalization issues
- Offline RL and Policy constraints
Why you should choose this course about DRL?
- DRL Is a very promising type of learning as it does not need to know the solution
- DRL Only needs the rules and good rewards
- DRL Combines the best aspects of deep learning and reinforcement learning.
- DRL has achieved impressive results in games, robotic, finance and many more fields
References
- Bertsekas, Dynamic Programming and Optimal Control, Vols I and II
- Goodfellow, Bengio, Deep Learning
- Powell, Approximate Dynamic Programming
- Puterman, Markov Decision Processes: Discrete Stochastic Dynamic Programming
- Sutton & Barto, Reinforcement Learning: An Introduction
- Szepesvari, Algorithms for Reinforcement Learning
Learning outcomes :
What you will acquire in this class?
- Understand principles of Deep Reinforcement Learning (DRL)
- Know main DRL algorithms
- Get some intuition about what DRL is good and not good at?
- Program DRL algorithms
-
Anonymisation, confidentialité
Anonymisation, confidentialité
Ects : 3
Lecturer :
Pierre SENELLART
Total hours : 24
-
Data wrangling, qualité de données
Data wrangling, qualité de données
Ects : 3
Lecturer :
Pierre SENELLART
Total hours : 24
-
Fondements des langages de requête graphe et RDF
Fondements des langages de requête graphe et RDF
Ects : 3
Total hours : 24
-
Semaine intensive PSL Humanité numérique
Semaine intensive PSL Humanité numérique
Ects : 3
-
Semaine intensive PSL Géonomique
Semaine intensive PSL Géonomique
Ects : 3
UE Obligatoires
-
Stage
Stage
Ects : 12
Academic Training Year 2023 - 2024 - subject to modification
Teaching Modalities
The program starts in September and attendance is required.
The year consists of a semester of advanced instruction in AI (from September to December) followed by a semester of electives (from January to April) and a research internship (April to September). Courses are divided over two semesters. During the first core semester, from September to Decembers, students must take six courses in artificial intelligence and data science, for a total of 30 ECTS credits. During the second, elective semester, from January to April, students must choose at least six advanced courses from a wide range of options (advanced machine learning, natural language processing, scatter plots and 3D modeling, ethics and data science, data mining, etc.) for a total of 18 ECTS credits.
For students who need it, a refresher course in foundations of mathematics and computer science is offered before the core semester begins in September.
UE obligatoires
-
Fondamentaux de l’apprentissage automatique
Fondamentaux de l’apprentissage automatique
Ects : 3
Lecturer :
GEOVANI RIZK
Overview :
The aim of this course is to provide the students with the fundamental concepts and tools for developing and analyzing machine learning algorithms.
The course will introduce the theoretical foundations of machine learning, review the most successful algorithms with their theoretical guarantees, and discuss their application in real world problems. The covered topics are:
- Introduction to the different paradigms of ML and applications
- Computational learning theory - PAC model - VC-dimension - Rademacher complexity,… - Supervised learning - Logistic regression and beyond - Perceptron - SVM - Kernel methods - Decision trees and Random Forests - Ensemble methods: bagging and boosting - Unsupervised learning - Dimensionality reduction: PCA, ICA, Random Projections, Kernel PCA, ISOMAP, LLE - Density estimation - EM - Spectral clustering - Online learning - Multiclass and ranking algorithms
- Practical sessions
Bibliography-recommended reading
References: - Mohri, M., Rostamizadeh, A., & Talwalkar, A. (2012). Foundations of machine learning. MIT press. - Shalev-Shwartz, S., & Ben-David, S. (2014). Understanding machine learning: From theory to algorithms. Cambridge university press. - Vapnik, V. (2013). The nature of statistical learning theory. Springer science & business media. - Bishop Ch. (2006). Pattern recognition and machine learning. Springer - Friedman, J., Hastie, T., & Tibshirani, R. (2001). The elements of statistical learning (Vol. 1, No. 10). New York, NY, USA:: Springer series in statistics. - James, G., Witten, D., Hastie, T., & Tibshirani, R. (2013). An introduction to statistical
-
Optimisation pour l'apprentissage automatique
Optimisation pour l'apprentissage automatique
Ects : 3
Lecturer :
Clement ROYER
Total hours : 18
Overview :
Optimization is at the heart of most recent advances in machine learning. Indeed, it not only plays a major role in linear regression, SVM and kernel methods, but it is also the key to the recent explosion of deep learning for supervised and unsupervised problems in imaging, vision and natural language processing. This course will review the mathematical foundations, the underlying algorithmic methods and showcase modern applications of a broad range of optimization techniques.
The course will be composed of classical lectures and numerical sessions in Python. It will begin with the basic components of smooth optimization (optimality conditions, gradient-type methods), then move to methods that are particularly relevant in a machine learning setting such as the celebrated stochastic gradient descent algorithm and its variants. More advanced algorithms related to non-smooth and constrained optimization, that encompass known characteristics of learning problems such as the presence of regularizing terms, will also be described. During lab sessions, the various algorithms studied during the lectures will be implemented and tested on real and synthetic datasets: these sessions will also address several practical features of optimization codes such as automatic differentiation, and built-in optimization routines within popular machine learning libraries such as PyTorch.
Recommended prerequisites :
Fundamentals of linear algebra and real analysis. Experience with Python programming.
Learning outcomes :
- Identify the characteristics of an optimization problem given its formulation.
- Know the theoretical and practical properties of the most popular optimization techniques.
- Find the best optimization algorithm to handle a particular feature of a machine learning problem.
Assessment :
Written exam.
Bibliography-recommended reading
Bibliography: Theory and algorithms: Convex Optimization, Boyd and Vandenberghe Introduction to matrix numerical analysis and optimization, Philippe Ciarlet Proximal algorithms, N. Parikh and S. Boyd Introduction to Nonlinear Optimization - Theory, Algorithms and Applications, Amir Beck Numerics: Pyrthon and Jupyter installation: use only Python 3 with Anaconda distribution. The Numerical Tours of Signal Processing, Gabriel Peyré Scikitlearn tutorial #1 and Scikitlearn tutorial #2, Fabian Pedregosa, Jake VanderPlas Reverse-mode automatic differentiation: a tutorial Convolutional Neural Networks for Visual Recognition Christopher Olah, Blog
-
Bases de données avancées (SBGD non classiques)
Bases de données avancées (SBGD non classiques)
Ects : 3
Total hours : 24
Overview :
Le cours a pour objectif d'apprendre aux étudiants les aspects fondamentaux des différents types bases de données qu’elles soient basées sur le SQL, le NoSQL (Not Only SQL) ou récemment le NewSQL.
Le cours s’articule en trois parties. Dans la première partie, l’accent est mis sur les bases de données relationnelles : leurs avantages et leurs inconvénients, ainsi que la correspondance objet-relationnel (Object-Relationnel Mapping -ORM) avec la norme JPA. La deuxième partie présentera les différents modèles noSQL (clé-valeur, document, graphe), les notions de disponibilité et de partitionnement à la cohérence (propriétés BASE, théorème CAP), les différents systèmes NoSQL (MongoDB, Cassandra, CouchBase, ...), les avantages et les inconvénients du NoSQL. La troisième partie sera consacrée aux bases NewSQL : leur définition et leurs caractéristiques, les nouvelles architectures et la notion de DBaaS (Database as a service), leurs avantages et leurs inconvénients. Les notions apprises seront mises en pratique dans un projet où les étudiants devront manipuler différents types de bases de données afin de les comparer.
-
Apprentissage Profond
Apprentissage Profond
Ects : 3
Lecturer :
TRISTAN CAZENAVE
Total hours : 24
Overview :
This course is about using deep learning tools.
The objective of the course is to be able to design deep neural networks and to apply them to various problems. The language used for the course is Torch. It relies on the Lua scipting language augmented with tensor specific instructions. During the course, we will use simple examples to learn how to generate and transform data in Torch as well as how to learn from this data. We will cover deep neural networks, deep convolutional neural networks and some optimizations of the architecture such as residual nets.
-
Systèmes, paradigmes et langages pour les Big Data
Systèmes, paradigmes et langages pour les Big Data
Ects : 3
Lecturer :
DARIO COLAZZO
Total hours : 24
Overview :
The main aim of this course is to give students a deep and solid understanding of the state of the art of Big Data systems and programming paradigms, and to enable them to devise and implement efficient algorithms for analysing massive data sets.
The focus will be on paradigms based on distribution and shared-nothing parallelism, which are crucial to enable the implementation of algorithms that can be run on clusters of computers, scale as the size of input data increases, and can be safely executed even in the presence of system failures.
Lectures will give articular emphasis to the MapReduce paradigm and the internal aspects of its related runtime support Hadoop, as well as to MapReduce-based systems, including Spark and Hive, that provide users with powerful programming tools and efficient execution support for performing operations related to complex data flows. The attention will be then given to mechanisms and algorithms for both iterative and interactive data processing in Spark. A particular attention will be given to SQL-like data querying, graph analysis, and the development of machine learning algorithms.
A large part of the course consists of lab-sessions where students develop parallel algorithms for data querying and analysis, including algorithms for relational database operators, matrix operations, graph analysis, and clustering. Lab-sessions rely on the use of both desktop computers and Hadoop clusters on the Amazon WS cloud.
Program:
1. Introduction to massive data management and processing.
2. A data operating system for distributed data management, Hadoop.
3. MapReduce paradigm, algorithm design, implementation and optimisation.
4. iterative and interactive massive data processing, algorithm design, implementation and optimisation in Spark
5. large scale data-warehouse in Hive
References:
Mining of Massive Datasets. Jure Leskovec, Anand Rajaraman, Jeff Ullman www.mmds.org
Data-In tensive Text Processing with MapReduce. Jimmy Lin and Chris Dyer. Mogan & Claypool Publishers
Hadoop: The Definitive Guide - Tom White. O'Reilly.
Apache Hadoop Yarn - Arun C.Murty, Vinod Kumar Vavilapalli, et al. Addison Wesley
Programming Hive. Edward Capriolo, Dean Wampler, Jason Rutherglen. O'Reilly.
Big Data Analytics with Spark.
-
Ethique et science des données
Ethique et science des données
Ects : 3
Lecturer :
OLIVIA TAMBOU
Total hours : 12
Overview :
The course will be the occasion, for future data scientists, and for students in general, to question the benefits and risks of science.
The course will permit them to approach from a pragmatic viewpoint questions they may have to face some day, and issues such as the various facets of privacy, the fairness of automatic decisions, the transparency of algorithmic processes, their explainability.
Learning outcomes :
Les étudiants devront être capable de comprendre les principaux enjeux éthiques en matière de données et d'intelligence artificielle tout en ayant un aperçu des différentes obligations juridiques existantes et en cours de création à l'échelle de l'Union européenne
-
Fouille de Graphes
Fouille de Graphes
Ects : 3
Lecturer :
MAIXENT CHENEBAUX
Total hours : 24
Overview :
The objective of this course course is to give students an overview of the field of graph mining and network science. Since graphs form a complex and expressive data type, we need methods for extracting information efficiently. Moreover, graph applications are very diverse and need specific algorithms.
The course presents new ways to model, mine and analyze graph-structured data and include many examples of applications. Lab sessions are included alowing students to practice graph mining and network science.
Outline of the course:
1. Centrality measures 2. Spectral graph theory and graph signal processing 3. Community detection 4. Machine learning and deep learning on graphs 5. Node classification and link prediction 6. Graph representation learning 7. Diffusion process and epidemics on networks
Learning outcomes :
1. Manipulate and create graphs using Python's NetworkX library
2. Master the centrality, community detection, classification and machine learning algorithms
3. Know how to use your knowledge in network science to solve problemes arising in other domains (cloud points, image, audio files, ...)
-
Data wrangling, qualité de données
Data wrangling, qualité de données
Ects : 3
Lecturer :
VERONIQUE FOURNEL
Total hours : 21
Overview :
Ce cours a pour objectif d’enseigner une méthodologie pour diagnostiquer et corriger les problèmes dus à la non qualité des données, mettre en œuvre une démarche qualité des données et mesurer ses effets. Il donne également un aperçu des outils existants et de leur utilisation.
Les différentes sources de données et leur exploitation. Mesure de la qualité des données et principales méthodes existantes. Cout de la qualité. Méthodes d’identification et de correction des données suivant leur type (manquantes, aberrantes, erronées, …). Indicateurs et suivi qualité des données. Amélioration de la qualité des données. Les outils logiciels et la qualité des données.
-
Traitement automatique des langues - NLP
Traitement automatique des langues - NLP
Ects : 3
-
Apprentissage par renforcement
Apprentissage par renforcement
Ects : 3
Lecturer :
STEPHANE AIRIAU
Total hours : 24
UE obligatoires
-
Apprentissage profond pour l’analyse d’images
Apprentissage profond pour l’analyse d’images
Ects : 3
Lecturer :
Thomas RICCATE
Total hours : 24
Overview :
Deep learning has achieved formidable results in the image analysis field in recent years, in many cases exceeding human performance. This success opens paths for new applications, entrepreneurship and research, while making the field very competitive.
This course aims at providing the students with the theoretical and practical basis for understanding and using deep learning for image analysis applications.
The course will be composed of lectures and practical sessions. Moreover, experts from industry will present practical applications of deep learning. Lectures will include: • Introduction to machine learning • Artificial neural networks, back-propagation algorithm • Convolutional neural network • Design and optimization of a neural architecture • Successful architectures (AlexNet, VGG, GoogLeNet, ResNet) • Analysis of neural network function • Image classification and segmentation • Auto-encoders and generative networks • Current research trends and perspectives
During the practical sessions, the students will code in Python, using Keras and Tensorflow. They will be confronted with the practical problems linked to deep learning: architecture design; optimization schemes and hyper-parameter selection; analysis of results.
Prerequisites: Linear algebra, basic probability and statistics
-
Flux de données
Flux de données
Ects : 3
Total hours : 24
Overview :
Ce cours a pour objectif de décrire les principes des systèmes capables de traiter les grandes masses de données en temps réel ou en temps quasi-réel et d’expliquer les apports des architectures microservices dans ce contexte.
Ce cours est découpé en trois parties : -Streaming des données : Présentation des différentes architectures et frameworks permettant de capturer, traiter, analyser et visualiser les données massives en temps réel -Architectures microservices : Principes de découpage des systèmes en services simples, facilement couplés assurant l’agilité du système global ainsi que les technologies et les pratiques de développement associés seront traités dans cette partie du cours. -Projet : Mise en pratique avec Java d’une application mettant en œuvre Spark Streaming et les microservices en REST.
-
Recherche Monte-Carlo et Jeux
Recherche Monte-Carlo et Jeux
Ects : 3
Lecturer :
TRISTAN CAZENAVE
Total hours : 24
Overview :
Ce cours est une introduction aux méthodes dites de Monte-Carlo. Ces méthodes sont utilisées pour calculer des espérances, et par extension, des intégrales par simulation. L’objectif de ce cours est non seulement de fournir les bases théoriques des méthodes de Monte-Carlo, mais aussi de fournir les outils permettant leur utilisation pratique à travers des TP.
Le cours couvre les sujets suivants : -introduction de la méthode de Monte-Carlo -techniques de réduction de variance -introduction aux suites à discrétion faible
-
Visualisation de données
Visualisation de données
Ects : 3
Total hours : 24
Overview :
Ce cours a pour objectif de décrire les démarches, méthodes et outils utilisés pour représenter les données complexes et multiples issues des grandes masses de données en visuels simples à comprendre et à interpréter, notamment pour les utilisateurs métier et pour les décideurs.
Le cas de l’apport de la visualisation des données sous différentes formes graphiques lors de la préparation des données en amont de la modélisation et de l’utilisation des modèles et algorithmes de Machine Learning sera développé. Le cours s’appuie sur de nombreux exemples puisés dans les domaines de la finance, de la santé, du marketing et des travaux pratiques sur des cas concrets sont également prévus.
-
IA sur le Cloud
IA sur le Cloud
Ects : 3
Total hours : 24
Overview :
The main aim of this course is to present to students and give them the possibility to acquire knowledge about typical Cloud architectures to support all the phases of typical IA data processing:
Covered topics include data storage and preparation as well as deployment and execution of machine learning algorithms. A particular attention will be given to the typical cloud architectures and the way they can ensure optimal data processing in IA pipelines, by taking into account the monetary cost of resources among other traditional parameters.
-
Graphes de connaissance, logiques de description, raisonnement sur les données
Graphes de connaissance, logiques de description, raisonnement sur les données
Ects : 3
Lecturer :
Michael THOMAZZO
Camille BOURGAUX
Total hours : 24
Overview :
Introduction to Knowledge Graphs, Description Logics and Reasoning on Data.
Knowledge graphs are a flexible tool to represent knowledge about the real world. After presenting some of the existing knowledge graphs (such as DBPedia, Wikidata or Yago) , we focus on their interaction with semantics, which is formalized through the use of so-called ontologies. We then present some central logical formalism used to express ontologies, such as Description Logics and Existential Rules. A large part of the course will be devoted to study the associated reasoning tasks, with a particular focus on querying a knowledge graph through an ontology. Both theoretical aspects (such as the tradeoff between the expressivity of the ontology language versus the complexity of the reasoning tasks) and practical ones (efficient algorithms) will be considered.
Program: 1. Knowledge Graphs 2. Ontology Languages (Description Logics, Existential Rules) 3. Reasoning Tasks (consistency, classification, ontological query answering) 4. Ontological Query Answering (forward and backward chaining, decidability and complexity, algorithms, advanced topics)
References: -- The description logic handbook: theory, implementation, and applications. Baader et al., Cambridge University Press -- Foundations of Semantic Web Technologies, Hitzler et al., Chapman&Hall/CRC -- Web Data Management, Abiteboul et al., Cambridge University Press
Prerequisites: -- first-order logic; -- complexity (Turing machines, classical complexity classes) is a plus.
Learning outcomes :
Understanding of the problems studied in the course.
Assessment :
Examen
-
Machine Learning sur Big Data
Machine Learning sur Big Data
Ects : 3
Total hours : 24
Overview :
This course focuses on the typical, fundamental aspects that need to be dealt with in the design of machine learning algorithms that can be executed in a distributed fashion, on Hadoop clusters, in order to deal with big data sets, by taking into account scalability and robustness.
Machine learning algorithms are more and more used today, and there is an ever increasing demand of machine learning algorithms that scales over massives data sets. This course focuses on the typical, fundamental aspects that need to be dealt with in the design of machine learning algorithms that can be executed in a distributed fashion, on Hadoop clusters, in order to deal with big data sets, by taking into account scalability and robustness. So the course will focus on a bunch of main-stream, sequential machine learning algorithms, by taking into account the following crucial and complex aspects. The first one is the re-design of algorithms by relying on programming paradigms for distribution and parallelism based on map-reduce, to this end Spark will be used. The second aspect is experimental analysis of the Spark implementation of designed algorithms in order to test their scalability and precision. The third aspect concerns the study and application of optimisation techniques in order to overcome lack of scalability and to improve execution time of designed algorithm.
The attention will be on machine learning technique for dimension reduction, clustering and classification, whose underlying implementation techniques are transversal and find application in a wide range of machine learning algorithms. For some of the studied algorithms, the course will present techniques for a from-scratch implementation in Spark core, while for other algorithms Spark ML will be used and end-to-end pipelines will be designed. In both cases algorithms will be analysed and optimised on real life data sets, by relaying on a local Hadoop cluster, as well as on a cluster on the Amazon WS cloud.
Bibliography-recommended reading
References: - Mining of Massive Datasets www.mmds.org - High Performance Spark - Best Practices for Scaling and Optimizing Apache Spark Holden Karau, Rachel Warren O'Reilly
-
Projet Sciences des Données
Projet Sciences des Données
Ects : 3
Lecturer :
GEOVANI RIZK
Total hours : 24
Overview :
The goal of this module is to provide students with a hands-on experience on a novel data-science/AI challenge using state-of-the-art tools and techniques discussed during other classes of this master.
Students enrolled in this class will form groups and choose one topic among a list of proposed topics in the core areas of the master such as supervised or unsupervised learning, recommendation, game AI, distributed or parallel data-science, etc. The topics will generally consist in applying a well-established technique on a novel data-science challenge or in applying recent research results on a classical data-science challenge. Either way, each topic will come with its own novel scientific challenge to address. At the end of the module, the students will give an oral presentation to demonstrate their methodology and their findings. Strong scientific rigor as well as very good engineering and communication skills will be necessary to complete this module successfully.
UE obligatoires
-
Memoire
Memoire
Ects : 9
Academic Training Year 2023 - 2024 - subject to modification
Teaching Modalities
The program starts in September and attendance is required. Students rotate between four weeks at a company and two or three weeks at the university.
The program is divided into two semesters, S3 and S4. Each semester is made up of a course, as well as a thesis in S4.
The final grade for a course is the cumulation of grades for continuous assessment, projects, homework, oral or written exams, and participation. Every course for which a student receives a final grade of 10/20 or above is deemed passed, and the appropriate ECTS credits are granted.
Each semester is made up of courses, as well as a thesis in S4. A student will have passed a semester if all the following conditions are met:
- They take at least 30 ECTS Their final grade for the semester is at least 10/20
- Their final grade for the semester is at least 10/20.
- The final grade for each course taken that semester is at least 6/20
- The final grade for the thesis in semester 4 is at least 10/20
- If a student has passed a semester, they are to considered to have passed all the courses that make up that semester and to have received the associated ECTS credits.
To pass a year, the student must have passed both semesters and all the courses that make up that unit and to have received the associated ECTS credits. A student will have passed a year (and received the associated 60 ECTS credits) if all the following conditions are met:
- They have taken at least 60 ECTS credits and receive a final grade for the year of at least 10/20
- The final grade for each semester is at least 10/20
- The final grade for each course in each semester is at least 6/20
- The final grade for the thesis in semester 4 is at least 10/20
Internships and Supervised Projects
Students intern in an academic or private research center from April to August and then write and defend a thesis in September.
Research-driven Programs
Training courses are developed in close collaboration with Dauphine's world-class research programs, which ensure high standards and innovation.
Research is organized around 6 disciplines all centered on the sciences of organizations and decision making.
Learn more about research at Dauphine