Play game 2048 with an "artificial intelligence"

IA created :

• Random move. Average score : ~200 pts
• Maximize score at next move. Average score : ~500 pts
• Monte carlo simulations, with parallel simulations. 5000 simulation, 6 nodes deep : up to 70.000 pts
• Minimax : max > 120.000 pts
• Expectimax
• Reinforcement learning : Q-learning

Next steps :

• Use C++ / Cython to accelelerate MC simulations
• Add Alpha–beta pruning to minimax algo
• Use bandit algorithms for MC simulations : UCB, UCT
• Machine learning