Quantitative Methods in AI (QMAI)

March 31 to April 2, 2021

Fees: 1200 euros
Preferential fees for UGA/CNRS.


The development of Artificial Intelligence (AI) is nowadays omnipresent in various industrial solutions. The objective of this training is to discover the basic principles as well as the main quantitative algorithms in AI.


Session 1

  • Lecture: Supervised classification (4h)
- Principle of induction in AI ;
- Basic assumptions and paradigms of quantitative methods in AI ;
- Principle of empirical risk minimization ;
- Perceptron algorithm.
  • Practice: (2h)
- Environnement Jupyter/Python ;
- Implementation of perceptron in Python (2h).

Session 2

  • Lecture: supervised classification (continuation - 4h)
- Gradient descent algorithm, Adaptive Linear Neuron (Adaline) ;
- Comparison between Perceptron and Adaline.
  • Practice:
    - Implementation of Adaline in Python (2h).

Session 3

  • Lecture: Unsupervised learning (4h)
- Basic paradigm ;
- Maximization of complete Maximum Likelihood ;
- K-means algorithm.
  • Practice:
    - Implementation of Kmeans in Python (2h).
Facilitation Massih-Reza Amini, Romain Couillet, Franck Iutzeler
Public Engineers
Volume The training includes 3 sessions of 6 hours: divided into 4 hours of theoretical lectures and 2 hours of implementation in Python.
Participants 6 to 12

Training program