SUMMARY
By adopting reinforcement learning to play the game, researchers have typically applied deep neural_networks to approximate the agent`s policy, mapping from game states to actions. The 2048 game has been used as a benchmark and testing ground for various AI algorithms including but not limited to reinforcement learning, neural_networks, evolutionary algorithms, and search algorithms. In the study by Kondo et_al, a convolutional neural_network (CNN) was trained to play 2048. A CNN is a type of neural_network that is particularly effective at processing images and other multidimensional data. CNN was able to achieve high scores in 2048 . . .
If you want to have access to all the content you need to log in!
Thanks :)
If you don't have an account, you can create one here.