The amazing part of what DeepMind has achieved is its capability to learn from raw pixel input with deep convolutional neural networks, which as I understand it quite different from what you do.

Still the reinforcement learning part is the same, but reinforcement learning was not the main contribution of this nature paper.