Course M2 IASD, Fondamentaux de l’apprentissage automatique

Master 2 students, Universite Paris, 1900

Information for the students:

The room for question-answer sessions and lectures: B113

Schedule for the lectures/TDs:

Lecture ‘Introduction and basic concepts’ (Total duration: 4 hours)
1.1. Introductory words (30 mins)
1.2. Learning machine learning (22 mins)
1.3. Data space (22 mins)
1.4. Distance in data space (42 mins)
1.5. Mean point (19 mins)
1.6. Conditional probability (34 mins)
1.7. Bayesian networks (39 mins)
1.8. Probability Density Function (35 mins)
Lecture ‘Clustering’ (Total duration 2.5 hours)
2.1. Introduction of clustering problem (25 mins)
2.2. K-means clustering algorithm (33 mins)
2.3. Hierarchical clustering (23 mins)
2.4. Density- and graph-based clustering (29 mins)
2.5. Assessment of clustering quality (31 mins)
Lecture ‘Dimensionality reduction’ (Total duration 3 hours)
3.1. Introduction to dimensionality reduction (35 mins)
3.2. Principal Component Analysis (64 mins)
3.3. Linear latent factor methods: ICA, NMF, FA (70 mins)
3.4. Multidimensional scaling (20 mins)
Lecture ‘Manifold learning’ (3 hours)
4.1. Introduction to manifold learning (20 mins)
4.2. Methods with explicit manifold: SOM, Principal Curves, manifolds and graphs (70 mins)
4.3. t-SNE and UMAP (50 mins)
4.4. Neural network-based autoencoders (30 mins)
Lecture ‘Selected topics in unsupervised machine learning’ (coming soon)
5.1. Basic notions on high-dimensional geometry
5.2. Estimating the intrinsic dimensionality of data
5.3. Basic notions on information geometry
5.4. Basic notions on optimal transport