Coding Sessions

Coding skills are important for successful AI research to implement algorithms, conduct experiments, analyze data, etc. To bridge the gap between abstract formulas and executable code, I occasionally organize voluntary coding sessions as an onboarding opportunity for students and fellow scientists.

The goal is to develop a basic understanding about the implementation details of planning and reinforcement algorithms. A simple navigation domain is used to evaluate the implemented algorithms.

The sessions cover the following topics:

1. Introduction to Decision Making
2. Monte Carlo Planning (Random Search, MCTS)
3. Reinforcement Learning (Q-Learning, DQN)
4. Advanced Topics (Policy Gradients, AlphaZero)

Check out the code base here.