These theories are before 2013.New theories could be found in deeplearning webset.
1.Feed-Forward propagation is a simple architecture of neural nets.
2.Above all are some calculation functions and models you might need or you will be familiar with.Here are some usual types of neurons,and from here, we will stand on the point of professor Hinton.
3.This one is just like a half of SVM，in professor Hinton‘s opinion，SVM is just a clever reincarnation of Perceptrons(ie.a neuron).
4.Convolutional neural networks
5."momentum"method is recommended.
6.more details about RNN
7.LSTM is a effective way that is recommended to learn RNN.
Help you and help me.Thanks!Good luck!
转载请注明出处 www.datageekers.com 十分感谢！