Introduction au Machine Learning

Intervenant : Guillaume Obozinski

Guillaume Obozinski is a research faculty at Ecole des Ponts – ParisTech, in the Imagine team of the Laboratoire d’Informatique Gaspard Monge. He is also associate member of the Sierra Inria team, in the Laboratoire d’Informatique de l’Ecole Normale Supérieure. His primary line of research examines Machine Learning and its applications.

Détail du cours

Lecture 1:

  • Decision theory and supervised learning:
  • Loss function, risk, target function, regression and multiclass classification settings
  • as example, empirical risk minimization, hypothesis space, 
  • overfitting, regularization, ridge regression, risk decomposition, cross-validation

Lecture 2:

  • Binary classification, evaluation of classifiers (ROC,precision recall, confusion matrix),
  • plug-in classification, local averaging methods, perceptron, logistic regression,  
  • Lecture 3:
  • Support vector machines:
  • (geometry and optimization, hinge loss, kernel trick, RKHS, kernel regression.

Les commentaires sont clos.