Understanding the Elements of Artificial Intelligence
Artificial Intelligence is here to stay. The development of AI is speeding up on a daily basis. Only recently, Google’s DeepMind created the AI AlphaStar that secured a decisive victory against two grandmaster players of the game of StarCraft II. In a series of test matches they played, the algorithm won 5-0.
This victory is a decisive moment for artificial intelligence, as the game of StarCraft II is fundamentally more difficult than the other games where Deepmind’s algorithm already claimed victory. Video games such as Atari, Mario, Quake III or Dota II are a lot less complex than StarCraft II. It is therefore remarkable that the AI, using a deep neural network that is trained directly from raw game data by supervised learning and reinforcement learning, managed to rival and surpass the skill of professional players.
The game requires a combination of changing short-term and long-term goals, adaption to unexpected events and situations with imperfect information. To achieve victory, the developers had to find solutions in fields such as game theory, long term planning (since cause-and-effects are not instantaneous in the game) and real-time actions. The developers of behind AlphaStar believe that the techniques used in this AI can also be relevant ...
Read More on Datafloq
This victory is a decisive moment for artificial intelligence, as the game of StarCraft II is fundamentally more difficult than the other games where Deepmind’s algorithm already claimed victory. Video games such as Atari, Mario, Quake III or Dota II are a lot less complex than StarCraft II. It is therefore remarkable that the AI, using a deep neural network that is trained directly from raw game data by supervised learning and reinforcement learning, managed to rival and surpass the skill of professional players.
The game requires a combination of changing short-term and long-term goals, adaption to unexpected events and situations with imperfect information. To achieve victory, the developers had to find solutions in fields such as game theory, long term planning (since cause-and-effects are not instantaneous in the game) and real-time actions. The developers of behind AlphaStar believe that the techniques used in this AI can also be relevant ...
Read More on Datafloq
Comments
Post a Comment