This paper released a novel and helpful way of training really deep neural networks by pre-training one particular hidden layer at a time utilizing the unsupervised learning method for limited Boltzmann equipment. Develop the subsequent technology of programs employing synthetic intelligence