This course introduces the main predictive methods based on machine learning. The main objective is to provide students with the knowledge necessary to understand supervised and unsupervised learning methods.
Course outline:
- Introduction
- Introduction to Machine Learning
- Reminders of linear algebra
- Supervised Learning (parametric methods: Linear Regression, Batch Gradient Descent, Stochastic Gradient Descent)
- Supervised Learning
- Introduction to Kaggle
- Linear algebra and normal equation
- Code vectorization
- Classification methods
- Parametric methods (logistic and softmax regression)
- Classification and regression trees (CART)
- Advanced tree-based methods (Bagging, Random Forest, Boosting) and variable importance
- Neural networks
- Introduction to neural networks
- Training a neural network
- Deep learning in Python (keras)
- Convolutional neural networks (CNN)
- Introduction to CNN
- Training a CNN
- Examples with keras
- Unsupervised learning
- Complements on unsupervised learning (clustering versus density estimation)
- Clustering methods (k-mans, HCA, DBSCAN)
- Recent approaches in clustering (CUBT, LDA)
Please note that other methods will be taught in Ewen Gallic’s course.