This course aims to provide students with a solid theoretical and practical foundation in mathematical optimization techniques essential to the development and refinement of machine learning algorithms and artificial intelligence applications. Students will learn to analyze and implement optimization methods, including gradient-based algorithms, adaptive learning rate techniques (e.g., Adam, RMSProp), automatic differentiation, and backpropagation, while addressing critical training challenges such as vanishing and exploding gradients. The course also covers neural network initialization strategies, dimensionality reduction (PCA), density estimation, and support vector machines (SVM), along with both unconstrained and constrained optimization problems. By the end of the course, students will be equipped to apply these techniques to improve model performance and solve complex problems across various AI domains.

Temps présentiel : 30 heures

Charge de travail étudiant : 70 heures

Méthode(s) d'évaluation : Examen final, Examen partiel, Travail personnel