Seonglae ChoClever and beautiful idea than Back Propagation way of get gradient. Geoffrey Hinton expected this method to be how the brain updates itself, but it is said that this turned out not to be the case.
확률적 그래픽 모델로, 각 노드가 이진 값을 가지며, 에너지를 최소화하는 방식으로 학습
The learning algorithm for Boltzmann machines is beautifully elegant, it’s maybe hopeless in practice but it’s the thing I enjoyed most developing even if it’s wrong - Geoffrey Hinton
