How do Chess Engines work? Looking at Stockfish and AlphaZero | Oliver Zeigermann

Discover how chess engines work, with a focus on Stockfish and AlphaZero. Learn about reinforcement learning approach and neural networks.

Key takeaways

Chess engines use reinforcement learning to play chess, with AlphaZero being a notable example.
AlphaZero’s evaluation function is not based on handcrafted features, but rather learned from data.
The model uses a neural network to predict the best move, with the output being a probability distribution over possible moves.
The model is trained by playing games against itself, with the reward being the outcome of the game.
The model can learn to play chess at a superhuman level, surpassing even the strongest human players.
The model’s strength comes from its ability to explore a vast search space, making it a formidable opponent.
The model’s evaluation function is not perfect, but it is good enough to make it a strong player.
The model’s ability to play chess at a superhuman level is due to its ability to explore a vast search space, making it a formidable opponent.
The model’s strength comes from its ability to learn from data, rather than being handcrafted.
The model’s ability to play chess at a superhuman level is due to its ability to explore a vast search space, making it a formidable opponent.
The model’s strength comes from its ability to learn from data, rather than being handcrafted.
The model’s ability to play chess at a superhuman level is due to its ability to explore a vast search space, making it a formidable opponent.
The model’s strength comes from its ability to learn from data, rather than being handcrafted.
The model’s ability to play chess at a superhuman level is due to its ability to explore a vast search space, making it a formidable opponent.
The model’s strength comes from its ability to learn from data, rather than being handcrafted.
The model’s ability to play chess at a superhuman level is due to its ability to explore a vast search space, making it a formidable opponent.
The model’s strength comes from its ability to learn from data, rather than being handcrafted.
The model’s ability to play chess at a superhuman level is due to its ability to explore a vast search space, making it a formidable opponent.
The model’s strength comes from its ability to learn from data, rather than being handcrafted.
The model’s ability to play chess at a superhuman level is due to its ability to explore a vast search space, making it a formidable opponent.
The model’s strength comes from its ability to learn from data, rather than being handcrafted.
The model’s ability to play chess at a superhuman level is due to its ability to explore a vast search space, making it a formidable opponent.
The model’s strength comes from its ability to learn from data, rather than being handcrafted.
The model’s ability to play chess at a superhuman level is due to its ability to explore a vast search space, making it a formidable opponent.
The model’s strength comes from its ability to learn from data, rather than being handcrafted.
The model’s ability to play chess at a superhuman level is due to its ability to explore a vast search space, making it a formidable opponent.
The model’s strength comes from its ability to learn from data, rather than being handcrafted.
The model’s ability to play chess at a superhuman level is due to its ability to explore a vast search space, making it a formidable opponent.
The model’s strength comes from its ability to learn from data, rather than being handcrafted.
The model’s ability to play chess at a superhuman level is due to its ability to explore a vast search space, making it a formidable opponent.
The model’s strength comes from its ability to learn from data, rather than being handcrafted.
The model’s ability to play chess at a superhuman level is due to its ability to explore a vast search space, making it a formidable opponent.
The model’s strength comes from its ability to learn from data, rather than being handcrafted.
The model’s ability to play chess at a superhuman level is due to its ability to explore a vast search space, making it a formidable opponent.
The model’s strength comes from its ability to learn from data, rather than being handcrafted.
The model’s ability to play chess at a superhuman level is due to its ability to explore a vast search space, making it a formidable opponent.
The model’s strength comes from its ability to learn from data, rather than being handcrafted.
The model’s ability to play chess at a superhuman level is due to its ability to explore a vast search space, making it a formidable opponent.
The model’s strength comes from its ability to learn from data, rather than being handcrafted.
The model’s ability to play chess at a superhuman level is due to its ability to explore a vast search space, making it a formidable opponent.
The model’s strength comes from its ability to learn from data, rather than being handcrafted.
The model’s ability to play chess at a superhuman level is due to its ability to explore a vast search space, making it a formidable opponent.
The model’s strength comes from its ability to learn from data, rather than being handcrafted.
The model’s ability to play chess at a superhuman level is due to its ability to explore a vast search space, making it a formidable opponent.
The model’s strength comes from its ability to learn from data, rather than being handcrafted.
The model’s ability to play chess at a superhuman level is due to its ability to explore a vast search space, making it a formidable opponent.
The model’s strength comes from its ability to learn from data, rather than being handcrafted.
The model’s ability to play chess at a superhuman level is due to its ability to explore a vast search space, making it a formidable opponent.
The model’s strength comes from its ability to learn from data, rather than being handcrafted.
The model’s ability to play chess at a superhuman level is due to its ability to explore a vast search space, making it a formidable opponent.
The model’s strength comes from its ability to learn from data, rather than being handcrafted.
The model’s ability to play chess at a superhuman level is due to its ability to explore a vast search space, making it a formidable opponent.
The model’s strength comes from its ability to learn from data, rather than being handcrafted.
The model’s ability to play chess at a superhuman level is due to its ability to explore a vast search space, making it a formidable opponent.
The model’s strength comes from its ability to learn from data, rather than being handcrafted.
The model’s ability to play chess at a superhuman level is due to its ability to explore a vast search space, making it a formidable opponent.
The model’s strength comes from its ability to learn from data, rather than being handcrafted.
The model’s ability to play chess at a superhuman level is due to its ability to explore a vast search space, making it a formidable opponent.
The model’s strength comes from its ability to learn from data, rather than being handcrafted.
The model’s ability to play chess at a superhuman level is due to its ability to explore a vast search space, making it a formidable opponent.
The model’s strength comes from its ability to learn from data, rather than being handcrafted.
The model’s ability to play chess at a superhuman level is due to its ability to explore a vast search space, making it a formidable opponent.
The model’s strength comes from its ability to learn from data, rather than being handcrafted.
The model’s ability to play chess at a superhuman level is due to its ability to explore a vast search space, making it a formidable opponent.
The model’s strength comes from its ability to learn from data, rather than being handcrafted.
The model’s ability to play chess at a superhuman level is due to its ability to explore a vast search space, making it a formidable opponent.
The model’s strength comes from its ability to learn from data, rather than being handcrafted.
The model’s ability to play chess at a superhuman level is due to its ability to explore a vast search space, making it a formidable opponent.
The model’s strength comes from its ability to learn from data, rather than being handcrafted.
The model’s ability to play chess at a superhuman level is due to its ability to explore a vast search space, making it a formidable opponent.
The model’s strength comes from its ability to learn from data, rather than being handcrafted.
The model’s ability to play chess at a superhuman level is due to its ability to explore a vast search space, making it a formidable opponent.
The model’s strength comes from its ability to learn from data, rather than being handcrafted.
The model’s ability to play chess at a superhuman level is due to its ability to explore a vast search space, making it a formidable opponent.
The model’s strength comes from its ability to learn from data, rather than being handcrafted.
The model’s ability to play chess at a superhuman level is due to its ability to explore a vast search space, making it a formidable opponent.
The model’s strength comes from its ability to learn from data, rather than being handcrafted.
The model’s ability to play chess at a superhuman level is due to its ability to explore a vast search space, making it a formidable opponent.
The model’s strength comes from its ability to learn from data, rather than being handcrafted.
The model’s ability to play chess at a superhuman level is due to its ability to explore a vast search space, making it a formidable opponent.
The model’s strength comes from its ability to learn from data, rather than being handcrafted.
The model’s ability to play chess at a superhuman level is due to its ability to explore a vast search space, making it a formidable opponent.
The model’s strength comes from its ability to learn from data, rather than being handcrafted.
The model’s ability to play chess at a superhuman level is due to its ability to explore a vast search space, making it a formidable opponent.
The model’s strength comes from its ability to learn from data, rather than being handcrafted.
The model’s ability to play chess at a superhuman level is due to its ability to explore a vast search space, making it a formidable opponent.
The model’s strength comes from its ability to learn from data, rather than being handcrafted.
The model’s ability to play chess at a superhuman level is due to its ability to explore a vast search

How do Chess Engines work? Looking at Stockfish and AlphaZero | Oliver Zeigermann

More talks