This unit aims to consolidate the mathematical and algorithmic foundations essential for mastering artificial intelligence.

• Key concepts in probability and statistics

• Data analysis and processing techniques

• Optimization methods for machine learning and deep learning

• Acquisition of the tools required for modelling, evaluating, and implementing AI systems

This unit introduces students to the foundations of symbolic AI, focused on knowledge representation and automated logical reasoning.

• Representation models: ontologies, knowledge graphs, inference rules, logic programmes

• Handling uncertainty and imprecision in reasoning

• Design of intelligent systems capable of deducing, explaining, and inferring from heterogeneous data

This unit trains students in the fundamental and advanced principles of machine learning and reinforcement learning.

• Supervised, unsupervised, and reinforcement learning approaches

• Key algorithms: regression, SVM, decision trees, clustering, Q-learning, etc.

• Design, training, validation, and evaluation of models on various datasets

• Managing bias, overfitting, interpretability, and performance optimization

• Practical workshops using modern frameworks (Scikit-learn, TensorFlow, PyTorch)

• Introduction to the design of intelligent agents capable of learning from experience

This unit aims to equip students with mastery of deep learning architectures and their applications.

• Traditional models: CNN, ResNet, RNN, LSTM

• Recent models: Transformers, diffusion models, GNNs, GANs

• Designing pipelines for training, evaluating, and adapting models

• Diverse applications: image segmentation, classification, text/image/sound generation, multimodal analysis

• Hands-on projects: computer vision (segmentation, object detection, activity recognition) and natural language processing

This unit trains students in emerging machine learning paradigms and hybrid AI architectures combining symbolic reasoning and deep learning.

• Continual (incremental) learning and self-supervised learning

• Robustness, scalability, and interpretability of models (Explainable AI)

• Foundation models and their use to generate specialised models

• Techniques for adapting and optimizing pre-trained models: distillation, fine-tuning, compression, pruning

• Designing hybrid AI architectures for inference, capable of explaining and justifying their decisions in complex tasks: information retrieval, semantic description, event prediction, decision-making, and automated planning

This unit explores the interfaces between AI and robotics, training students to design autonomous, intelligent robotic systems.

• Optimization and synthesis of robotic controllers

• Improvement of multimodal perception systems

• Autonomous planning of trajectories and complex tasks

• Target applications: hands-on experimentation on real and simulated robotic platforms (ROS-compatible) with exoskeleton robots and mobile or humanoid companion robots for personal assistance

This unit trains students in software engineering approaches for designing and implementing agent-centric AI systems.

• Study of NoSQL databases and creation of automated MLOps pipelines

• Multi-level deployment: Edge, Cloud, and IoT

• Designing complex software architectures for AI systems, including RAG (Retrieval-Augmented Generation)

• Management of autonomous multi-agent systems

• Integration and orchestration of AI agents capable of leveraging tools, reasoning, planning, and persistent memory to autonomously and collaboratively perform complex tasks

This unit enables students to apply their AI skills through a supervised, innovative project. Projects are presented and defended before a monitoring committee, with the possibility of submitting the work to national and international conferences or congresses.

• Discovering the lifecycle of an innovative AI project: ideation, proof of concept, evaluation, integration, and transfer to real-world use

• Teamwork on real-world problems proposed by the LISSI laboratory and industrial or academic partners

• Developing interdisciplinary projects using agile methodologies (e.g., SCRUM)

• Mobilizing different approaches: symbolic AI, deep learning, hybrid AI, robotics

• Developing autonomy, creativity, project management, and the capacity to innovate