The first of my three introductory AI courses: Python foundations, the building blocks of a machine learning program, and classic algorithmic techniques such as linear programming, greedy algorithms, recursion and dynamic programming.

Type

Course (part 1 of 3)

Duration

60h - 15 blocs/sittings

Materials

6 slide decks (PDF)

Language

French

View the course on GitHub

What the course covers

The course starts with Python: syntax and indentation, typing and casting, lists and dictionaries, functions and parameters, code documentation, imports, compile-time and logic errors, exceptions, unit tests, and popular open-source modules such as NumPy and scikit-learn, with formative exercises along the way.

It then walks through a complete machine learning program: importing the data and splitting the sets, the model’s topology, the loss function, the optimizer and the metric, then training, evaluation, plotting the results, and saving and loading a model.

The second half turns to algorithms: linear programming (decision variables, objective function, constraints, solving with PuLP and graphically with Matplotlib), greedy algorithms and Big O notation through the knapsack problem, recursion with the factorial example, and dynamic programming with Fibonacci, including a benchmark against plain recursion and a look at its decision tree.

Structure

• Sessions 2–3: introduction to Python
• Sessions 4–6: introduction to AI
• Session 7: work period
• Session 8: midterm exam
• Sessions 9–10: linear programming
• Session 11: greedy algorithms and Big O
• Session 12: recursion
• Session 13: dynamic programming
• Session 14: work period
• Session 15: final exam

Technologies

Python, NumPy, scikit-learn, TensorFlow and Keras, Matplotlib, PuLP and Google Colab.

View the course on GitHub