Inference, planning and RL from scratch
Particle filters, Bayesian networks, Q-learning and REINFORCE in NumPy
AI and Reinforcement Learning courses, ETF Belgrade. Code
From-scratch Python implementations with experimental analysis:
- Local search: simulated annealing, genetic algorithms, particle swarm optimization.
- Probabilistic inference: variable elimination, rejection and Gibbs sampling in Bayesian networks, particle-filter tracking.
- Reinforcement learning: Q-value iteration, Q-learning, REINFORCE.