What is Darkforest?
Darkforest is a computer Go program developed by Facebook's AI Research team, based on deep learning techniques using a convolutional neural network. It combines these techniques with Monte Carlo tree search (MCTS), a method commonly seen in computer chess programs.
Darkforest improved move prediction for Go compared to earlier MCTS only systems by using a policy network to guide search. Public evaluations reported strong amateur to professional level play depending on the version and time period, though reported ranks vary across servers and evaluation setups.
The Darkforest Go engine is open source, and its source code and pre-trained models have been publicly available on GitHub. This has been beneficial to the community, as it allows other researchers and developers to use and build upon the work done by the Facebook AI Research team.
In a broader context, the term dark forest is also used metaphorically in discussions about AI and generative models. It refers to the idea of an expanding landscape of AI generated content, where models can produce text, images, and other media that are increasingly difficult to distinguish from human created content. This metaphor draws from the dark forest theory in Liu Cixin's science fiction, which describes a universe where civilizations remain silent and hidden to avoid detection and potential destruction.
How does darkforest use deep learning techniques?
Darkforest, a computer Go program developed by Facebook's AI Research team, uses deep learning techniques in combination with a convolutional neural network (CNN). The CNN is designed to predict the next k moves based on the current state of the game and each player's territory. The architecture of Darkforest's CNN is a 12-layer full convolutional network with a width of 384 nodes. Each convolutional layer is followed by a rectified linear unit (ReLU), a popular activation function for deep neural networks.
The deep learning portion of Darkforest involves training a deep policy convolutional neural network. This neural network takes in a Go position and outputs the best moves. The training process involves feeding the network with game positions and the corresponding best moves, allowing the network to learn and predict the best moves given a game position.
In addition to the deep learning techniques, Darkforest also uses Monte Carlo Tree Search (MCTS), a method commonly seen in computer chess programs. The combination of deep learning and MCTS allows Darkforest to effectively evaluate the game state and make strategic decisions.
Some Darkforest variants experimented with architectural changes and different search integrations to improve efficiency and strength. Reported innovations include deeper networks and tighter coupling between policy evaluation and tree search.
The source code and pre-trained models of Darkforest are publicly available on GitHub, allowing other researchers and developers to use and build upon the work done by the Facebook AI Research team.