First you train the hidden layers individually in an unsupervised fashion using autoencoders. Then you train a final softmax layer, and join the layers together ...
確定! 回上一頁