Practical work on the use of Genetic Algorithms to minimize functions and solve gym environments.
To code the algorithms, we used the Deap library, and gym was used for the simulations.
We first worked on minimizing 2D functions, then higher dimension functions.
Then, we used these algorithms to solve a gym environment called cartpole, by learning an adapted policy.
Finally, we solved a more advanced version of this environment, using multi-objectives algorithms.
Note that the report (and the observations were made in French).