Human-level control through deep reinforcement learning

Demis Hassabis and colleagues at DeepMind have published their work on Q-networks. The agent is capable of learning how to play a number of Atari games by receiving only pixels and score as 'sensory input'. Its performance surpasses that of any professional human game tester.

Link to the article.

#machinelearning #artificialintelligence #deeplearning