Machine statistical Learning - 5PMSAST6

Goal(s)

Introduction to the statistical learning theory and prediction (regression/classification)

• Review of Models/Algorithms for supervised/unsupervised learning
• Illustration of these algorithms on different dataset
  (Artificial Intelligence, Bioinformatics, vision, etc ...)

Content(s)

• General introduction to the statistical learning theory and prediction (regression/classification)
• Generative approaches: Gaussian discriminant analysis, naïve Bayes hypothesis
• Discriminative approaches: linear models and penalization, stochastic gradient
• Model-free approaches: support vector machines (SVM)
• Model-free approaches: decision trees and random forests
• Unsupervised classification (kmeans, Gaussian mixture model, hierarchical methods)

• Basics of linear algebra
• Notions of probability (random vectors, Gaussian vectors, conditional distributions)
• Data analysis course: principal component analysis and dimension reduction
• Advanced notions of statistics (inference and statistical decision theory)
• Optimization: constrained optimization (KKT conditions), steepest descent algorithm