Differences

This shows you the differences between two versions of the page.

--- curriculum [2025/10/29 12:27] – director-llga
+++ curriculum [2025/10/29 15:06] (current) – director-llga
@@ Line 3: / Line 3: @@
 ===== MASTER 1  (M1) =====
-Mandatory:
+Two refresher courses:
   * Refresher in Statistics (APM_51438_EP, Marine Le Morvan, Inria)
   * Refresher in Computer Science (CSC_51440_EP, Amal Dev Parakkat, Telecom Paris)
@@ Line 15: / Line 15: @@
   * Machine Learning (MDC_51006_EP, Erwan Le Pennec and Jesse Reed, EP)
-course among:
+courses among:
-  * Deep Learning (Recommended option, CSC_51054_EP, Michalis Vazirgiannis and Johannes Lutzeyer, EP)
+  * Deep Learning (Recommended option, CSC_51054_EP, Michalis Vazirgiannis, Johannes Lutzeyer and Ye Zhu, EP)
   * Signal Processing (Recommended option, APM_51055_EP, Rémi Flamary, EP)
   * Emerging Subjects in Machine Learning and Collaborative Learning (Recommended option, APM_51178_EP, Aymeric Dieleveut and El Mahdi El Mhamdi, EP)
@@ Line 38: / Line 38: @@
 Mandatory:
   * Reinforcement Learning and Autonomous Agents  (CSC_52081_EP, Jesse Read, EP)
-  * Multimodal Generative AI (CSC_52002_EP, Vicky Kalogeiton, EP)
+  * Graph Machine and Deep Learning for Generative AI (CSC_52072_EP, Johannes Lutzeyer and Michalis Vazigiannis, EP)
+  * Introduction to Text Mining and NLP (CSC_52082_EP, Michalis Vazirgiannis and Davide Buscaldi, EP & Université Sorbonne Paris Nord)
-courses among:
-  * Deep Learning (Recommended option, APM_52183_EP, Kevin Scaman)
+course among:
-  * Image Synthesis: Theory and Practice (Recommended option, CSC_52084_EP, Tamy Boubekeur, Telecom ParisTech)
+  * Multimodal Generative AI (Recommended option, CSC_52002_EP, Vicky Kalogeiton, EP)
   * Statistics in Action (Recommended option, APM_52066_EP, Zacharie Naulet, EP & INRAE)
-  * Advanced Deep Learning (CSC_52087_EP, Michalis Vazirgiannis, Vicky Kalogeiton, Johannes Lutzeyer, EP)
+  * Deep Learning (Recommended option, APM_52183_EP, Kevin Scaman)
-  * Graph Representation Learning (CSC_52072_EP, Johannes Lutzeyer and Michalis Vazigiannis, EP)
+  * Computational Optimal Transport for ML and Fairness (Recommended option, Rémi Flamary, EP)
-  * Emerging Topics in Machine Learning (APM_52188_EP, Rémi Flamary, EP)
+  * Optimization for AI (APM_52067_EP, Luiz Chamon and Aymeric Dieuleveut, EP)
+  * Optimization and Responsible AI for Sustainability (CSC_52073_EP, Sonia Vanier, EP)
+  * Advanced Deep Learning (CSC_52087_EP, Vicky Kalogeiton, Johannes Lutzeyer, Ye Zhu and Xi Wang, EP)
+  * Image Synthesis: Theory and Practice (CSC_52084_EP, Tamy Boubekeur, Telecom ParisTech)
+  * Social Media and Communication: Probabilistic Models and Algorithms (APM_51058_EP, Laurent Massoulie, Inria)
 + Mandatory non-scientific courses:
@@ Line 64: / Line 70: @@
 ===== MASTER 2 (M2)======
-No refresher course is provided but students directly entering in the M2 and lacking background in Computer Graphics are welcome to follow the M1 refresher course in Computer Science.
+No refresher course is provided but students directly entering in the M2 and lacking background in language and graph modeling are welcome to follow the M1 refresher course in Computer Science.
+The detailed curriculum for M2 is currently under development. However, below we provide a preliminary overview of the core and introductory courses that students are expected to take.
 ==== Period 1 ====
-**APM_53440_EP - Advanced unsupervised learning (24h, 2 ECTS), Pierre Latouche(UCA)** (contact: pierre.latouche@uca.fr)
-This course aims at presenting advanced models and methods from computational statistics and machine learning for unsupervised learning, through the context of directed graphical models. Both the frequentist and Bayesian frameworks will be covered. In particular, we will study clustering methods and we will focus on mixture models and on the expectation maximisation algorithm for inference. The standard model selection criterion will be derived and we will illustrate their use to estimate the model complexity. After having given the theory of directed graphical models, we will show on a series of examples how the classical models from machine learning can be characterized in such a context. We will show how all the standard loss functions in machine learning can be linked to specific directed graphical models, giving rise to new ways to think about problems and to solve them. We will particularly insist on the interest of relying on directed graphical models to evaluate the complexity of the inference task for various models. The variational framework will be presented along with the variational EM and variational Bayes EM algorithms. The inference task of the stochastic block model, with a two-to-one dependency, will be described. The second part of the course will be about the use of deep neural networks in directed graphical models to obtain the so-called deep graphical models. Variational autoencoders and GAN we will be presented in such a context along with the corresponding inference strategies. We will finally show on a given example how the main model for social network analysis can be rewritten and extented with deep graphical models. We will point out the modifications in terms of directed graphical models, inference, and applications. In particular, we will describe the use of graph neural networks. Lectures will be done through slides and proofs on the blackboard. In labs, we will implement and test all the approaches seen in the lectures.
-**APM_53441_EP - From Boosting to Foundation Models: learning with Tabular Data (24h, 2 ECTS), Marine Le Morvan** (contact: marine.le-morvan@polytechnique.edu)
-> This course offers a modern overview of statistical learning with tabular data, from classical tree-based models to emerging deep and foundation models. We will review the foundations of gradient boosting methods and their scalable implementations, cover recent deep learning models tailored for tabular data, and introduce tabular foundation models. In particular, we will discuss the limitations of LLMs on structured data, introduce the concept of in-context learning, and provide an in-depth understanding of novel tabular foundation models, including their architecture and pretraining strategies. The course will also address key practical challenges in real-world datasets and applications, such as encoding heterogeneous feature types (categorical, numerical, temporal, textual), strategies for handling missing data, and methods for evaluating and calibrating predictive uncertainty.
 **CSC_53432_EP - Large Language Models (24h, 2 ECTS), Guokan Shang (MBZUAI) ** (contact: guokan.shang@mbzuai.ac.ae)
@@ Line 85: / Line 81: @@
 > The curriculum begins with a brief overview of key historical NLP techniques. It then transitions to the transformer architecture, focusing on its attention mechanism and tokenization—the core of modern LLMs. Pre-training objectives such as masked/denoising language modeling and causal language modeling will also be covered, forming the basis for models like BERT, GPT, and T5. The course then examines LLM post-training techniques used to refine pre-trained models, including instruction tuning (SFT), reinforcement learning from human feedback (e.g., PPO/DPO), and reinforcement learning from verifiable rewards (e.g., GRPO). Finally, the course will address LLM application and future directions—including RAG, agents, multimodality, and alternative model architectures.
-**CSC_53439_EP - Deep Reinforcement Learning (24h, 2 ECTS), Jesse Read** (contact: jesse.read@polytechnique.edu)
-> Reinforcement learning (RL) is of increasing relevance today, including in games, complex energy systems, recommendation engines, finance, logistics, and for auto-tuning the parameters of other learning frameworks. This course assumes familiarity with the foundations of RL and its main paradigms (temporal-difference learning, Monte Carlo, and policy-gradient methods). We will explore them further, and study modern state-of-the-art variants (such as proximal policy optimization), with a focus on developing RL solutions with deep neural architectures suited to modern applications. We will also take a look at specialized topics such inverse reinforcement learning.
 **CSC_53431_EP - Analysis and Deep Learning on Geometric Data (24h, 2 ECTS), Maks Ovsjanikov (EP)** (contact: maks@lix.polytechnique.fr)
 > This course will introduce students to advanced topics in modern geometric data analysis with focus on a) mathematical foundations (discrete differential geometry, mapping, optimization), and b) deep learning for best performing methods. We will give an overview of the foundations in shape analysis and processing before moving to modern techniques based on deep learning for solving problems such as shape classification, correspondence, parametrization, etc.
-**CSC_53433_EP - Creative & Generative models in Computer Graphics (24h, 2 ECTS),  Marie-Paule Cani (EP), Julien Pettré (Inria)** (contact: marie-paule.cani@polytechnique.edu)
-> This course presents the AI-related methods developed in Computer Graphics to create or generate individual 3D shapes, animated landscapes and humanoid motion. We first introduce user-centered Creative AI, i.e. smart 3D models - either based on knowledge or trained from examples, designed to help users creating and controlling 3D shapes and environments. Second, we focus on the use of AI in Character Animation, from early motion planning and control methods to deep reinforcement learning solutions. These methods result into 3D character models able to navigate alone or within crowds, and to interact with their environment. The lab sessions are held on Unity, based on C#.
 ==== Period 2 ====
-**CSC_54456_EP - Navigation for Autonomous systems (24h, 2 ECTS), David Filliat (ENSTA)** (contact: david.filliat@ensta-paris.fr)
+**To be completed soon.**
-> Drones and robots must create maps of their surroundings to plan their movement and navigate. This course presents the robotic platforms and the most common sensors (vision, Lidar, intertial units, odometry …) and the different components of navigation: control; obstacle avoidance; localization; mapping (SLAM) and trajectory planning as well as filtering (Kalman filter, particle filtering, etc.) and optimization techniques used in these fields.
-**CSC_54441_EP - Introduction to the verification of neural networks (24h, 2 ECTS), Eric Goubault (EP), Sylvie Putot (EP)** (contact: sylvie.putot@polytechnique.edu) - Elective course
-> Neural networks are widely used in numerous applications including safety-critical ones such as control and planning for autonomous systems. A central question is how to verify that they are correct with respect to some specification. Beyond correctness or robustness, we are also interested in questions such as explainability and fairness, that can in turn be specified as formal verification problems. In this course, we will see how formal methods approaches introduced in the context of program verification can be leveraged to address the verification of neural networks.
-**CSC_54445_EP - Geometric Algorithms for Point Patterns and 2D embedded Structures, Across Applications in Visual Computing (24h, 2 ECTS), Pooran Memari** (contact: pooran.memari@polytechnique.edu) - Elective course
-> This course explores the key role of 2D embedded structures, namely discrete distributions or point patterns in practical applications, such as object placement and texture synthesis. Through interconnected themes in data analysis, we will revisit foundational tools and introduce techniques to analyze, synthesize, design and further edit such 2D configurations. The exercise sessions present scenarios where geometric intuition leads to efficient computational methods, mainly via example-based modeling and lightweight learning techniques.
-**CSC_54442_EP - Socio-emotional embodied conversational agents (24h, 2 ECTS), Chloé Clavel (Inria Paris) and Brian Ravenet (Université Paris Saclay)** (contacts: chloe.clavel@inria.fr and brian.ravenet@lisn.upsaclay.fr>)
-> Many interactive systems, from virtual companions to online retailing, rely on embodied conversational agents. These agents need to reach a good level of communication skills to conduct a conversation with humans and be acceptable and trustworthy by humans. This course will introduce non-verbal behavior models, present models for multimodal dialog, opinion detection and voice quality, explain how to model the agent's emotions and their evolution over time, and present methods for enhancing naturalism with expressive gaze and gestures, realistic animation.
-**CSC_54444_EP - Virtual/Augmented Reality & 3D Interactions  (24h, 2 ECTS), Anatole Lécuyer (Inria Rennes), Ferran Argelaguet (Inria Rennes), Arnaud Prouzeau (Inria Saclay), Claudio Pacchierotti (CNRS - IRISA), Fabien Lotte (Inria Bordeaux)** (contact: anatole.lecuyer@inria.fr)
-> Metaverse and virtual/augmented reality technologies are spreading widely. But reconstructing our world and generating virtual ones would be useless without effective techniques to navigate and interact with them. This course will present virtual and augmented reality systems as well as the associated methods for 3D interaction, from multi-modal interaction merging visual immersion, sound and haptics systems to brain-computer interfaces.
-**CSC_54434_EP - 3D Computer Vision (24h, 2 ECTS), Xi Wang (EP)** (contact: Xi.Wang@polytechnique.edu)
-> This course presents modern 3D computer vision in a clear, step-by-step progression: we begin with classical multi-view reconstruction and structure‑from‑motion pipelines, then advance to neural implicit representations for novel‑view synthesis (e.g., NeRF), proceed to explicit geometry rendering via 3D Gaussian Splatting (3DGS), and finally explore generative 3D models: e.g., 3D generation guided by 2D generative score distillation (DreamFusion-like), data‑driven 3D/4D content generation for dynamic scenes and motion, and the latest video generation techniques.
-**CSC_54443_EP - Soft robots: Design, Modeling, Simulation and Control (24h, 2 ECTS),  Christian Duriez (Inria Lille)** (contact: christian.duriez@inria.fr)
-> Soft robotics is a promising novel field, bringing more robustness in robots design and for all tasks involving close interactions with humans, from help to disable people to medical robot. This course will give an introduction to recent advances in soft robotics, including design, modeling, simulation and control techniques for robots, and will present recent applications in medicine, industry and art.