Skip to main content
. 2015 Mar 3;10(3):e0115620. doi: 10.1371/journal.pone.0115620

Fig 5. Digit center reaching task.

Fig 5

(A) A set of digits used in the training. (B) The cumulative reward obtained with test dataset. (C, D) The activation of hidden neurons projected on the first two principal components in different reward settings. (C) The reward setting is the same as in the simple task. (D) The agent always gets reward of 2000 for any states and actions. Each point shows the hidden activation for each state using test digit dataset.