by Daniel Park
AlphaGo. Does that name sound familiar? Maybe, maybe not. Let’s go back to 1997, when AI (Artificial Intelligence) defeated the world chess champion. That is a huge achievement.
To begin with AlphaGo is a computer program, one designed to play the game of Go. Go is one of the most ancient board game still played today, which originated from China. Go may be even more complex and difficult than chess.
Basically, Go is a game with the objective of surrounding a larger total area of the board with one’s stones(game pieces) than the opponent. As the game progresses, the players position stones on the board to map out formations and potential territories.
Contests between opposing formations are often complex and may result in the expansion, reduction, or wholesale capture and loss of formation stones. Now what is so special about this AlphaGo?
On March 2016, AlphaGo beat a professional 9-dan (The highest ranking) Go player, Lee Sedol in a best of five series, winning four and losing only one. Okay, this may not sound like the most exciting news, but keep reading.
What’s interesting about this program (besides being the first computer program to beat a professional with no handicaps) is its mechanism. AlphaGo’s algorithm uses a combination of machine learning and tree-search techniques, combined with extensive training, both from human and computer play.
It uses a Monte Carlo tree search guided by a “value network” and a “policy network,” both implemented using deep neural network technology. A limited amount of game-specific feature detection pre-processing is applied to the input before it is sent to the neural networks.
The system’s neural networks were initially bootstrapped from human gameplay expertise. AlphaGo was initially trained to mimic human play by attempting to match the moves of expert players from recorded historical games, using a database of around 30 million moves.
Once it had reached a certain degree of proficiency, it was trained further by being set to play large numbers of games against other instances of itself, using reinforcement learning to improve its play.
Despite all the explanation of the technology of the AI, why was the victory so significant for AlphaGo?
The technology of AlphaGo is the core of future AI. Looking at the victory it may even redefine what we have thought of AI and future AI. Perhaps AlphaGo’s victory over one of the best Go players adds another milestone to the since 1997.