model_in_dqn

Find the explicit model in deep q network

Exp. 1 Supervised Maze (identical as VIN)

Current Results:

8x8: 99.02% (k=1,SAME/VALID,Batch_decay)
16x16: 97.18% (k=1,mpl_iter)
28x28: 95.07% (k=1,mpl_iter)