Introduction to Machine Learning with Python

This practice-oriented course aims to introduce participants to the world of machine learning, from mathematical basics, through simple regressions, to the perceptron model.

• Mathematical overview of 1-D linear regression
• Coding linear regression in Python
• Mathematics of multidimensional linear and polynomial regression
• Coding examples in Python
• Introduction to the classification problem
• Linear classifiers
• Biological motivation
• Logistic regression
• Feedforward mechanism and probabilistic interpretation
• Crossentropy error function
• Maximum-likelihood
• Gradient descent
• Presentation of practical problems: regularization, donut and XOR problems
• Backpropagation, or how a machine model learns
• From neurons to neural nets
• Softmax
• Building a complete deep neural network in Python
• Trenching
• Extracting predictions
• Revisiting practical problems: donut and XOR problem
• Hyperparameters and cross checking

The course requires completion (within 1 year) of Advanced Python Programming (PR-PYA) or equivalent (or equivalent) training. You must also have 1 year of daily Python programming experience.

Proficiency in English at document reading level is required to complete the training.