深度信念网络 (Deep Belief Network, DBN) 由 Geoffrey Hinton 在 2006 年提出。它是一种生成模型，通过训练其神经元间的权重，我们可以让整个神经网络按照最大概率来生成训练数据。我们不仅可以使用 DBN 识别特征、分类数据，还可以用它来生成数据。下面的图片展示的是用 DBN 识别手写数字： -Depth belief networks (Deep Belief Network, DBN) proposed by the Geoffrey Hinton in 2006. It is a model generated by heavy weight training nerve yuan between, we can make the entire neural network in accordance with the highest probability to generate training data. Not only can we use the DBN recognition feature, classification data, you can also use it to generate data. The following picture shows with DBN recognition of handwritten numbers: