UE Machine learning fundamentals

Degrees incorporating this pedagocial element :


  • Consistency of the Empirical Risk Minimization
  • Uniform Generalization Bounds and Structural Risk Minimization
  • Unconstrained Convex Optimization
  • Binary Classification algorithms (Perceptron, Adaboost, Logistic Regression, SVM) and their link with the ERM and the SRM principles
  • Multiclass classification
  • Application and experimentations

Evaluation : Homeworks (30%), Final exam (70%)


Statistics and probability (BSc)

Targeted skills

Understanding of fundamental notions in Machine Learning (inference, ERM and SRM principles, generalization bounds, classical learning models, unsupervised learning, semi-supervised learning.


[1] Massih-Reza Amini - Apprentissage Machine de la théorie à la pratique, Eyrolles, 2015.
[2] Christopher Bishop - Neural Networks for Pattern Recognition, Oxford University Press, 1995.
[3] Richard Duda, Peter Hart & David Strok - Pattern Classification, John Wiley & Sons, 1997.
[4] John Shawe-Taylor & Nello Cristianini - Kernel Methods for Pattern Analysis, Cambridge University Press, 2004.
[5] Colin McDiarmid - On the method of bounded differences,Surveys in Combinatorics, 141:148-188, 1989. 
[6] Mehryar Mohri, Afshin Rostamzadeh & Ameet Talwalker - Foundations of Machine Learning, MIT Press, 2012.
[7] Bernhard Schölkopf & Alexander J. Smola - Learning with Kernels, MIT Press, 2002.
[8] Vladimir Kolchinskii - Rademacher penalties and structural risk minimization, IEEE Transactions on Information Theory, 47(5):1902–1914, 2001.