Site Unistra - Accueil
Faire un don

Description

Machine learning and knowledge discovery from databases.

  • Understand machine learning
  • Overview of algorithms for clustering, classification, and association rule learning and focus on data representation
  • Practice with WEKA and KNIME softwares
  • Data pre-processing ; evaluation ; integration ; representations.
  • Frequent patterns and association rules.
  • Clustering : k means ; expectation maximization.
  • Classification : k nearest neighbours ; naive Bayesian classifier.
  • Decision trees : principle, classification, regression, sensitivity, random forest.
  • Neural networks : single and multiple layers ; backpropagation ; strengths and limits ; example (clustering of reactions by Kohonen maps).
  • Support Vector Machinees : principle, classification and regression.
  • Genetic algorithms : concepts ; fitness function ; crossover and mutations.
  • Labs with WEKA and KNIME.
  • Detailed examples

Compétences visées

  • Understand challenges and limits of machine learning
  • Choose relevant algorithms to cluster, classify or extract association rules from data
  • Application of those methods with WEKA and KNIME software

Contact

Responsable(s) de l'enseignement
Nicolas Lachiche : nicolas.lachiche@unistra.fr
Gilles Marcou : g.marcou@unistra.fr