Artificial neural network deep learning MLP RBF RBM DBN DBM CNN Finishing Learning

Note: Organize the PPT from shiming teacher

Content Summary 1 Development History2 Feedforward Network (single layer perceptron, multilayer perceptron, radial basis function network RBF) 3 Feedback Network (Hopfield network,Lenovo Storage Network, SOM,Boltzman and restricted Boltzmann machine rbm,dbn,cnn)

Development History




single-layer perceptron 1 Basic model
2 If the excitation function is linear, the least squares can be calculated directly 3 if the excitation function is sifmoid functions, the iterative update (one-time or sample-by-update) only makes a simple derivative expansion, it is easy to deduce




Multilayer Perceptron 1 Basic model

2 Example (Multilayer Perceptron MLP with one hidden layer)
Model:
Y=h (v) =h (H (U))
Solution: How is this converted to 6 K (xi)?

Then the weights of the two layers are then divided into two different values:
Then update, reverse propagate (BP)



3 Experience 4 Advantages and disadvantages


RBF Neural Network 1 Model 2 Solution
3 Advantages and perspectives




Introduction to Deep learning 1 Forward Neural network
2 Development History 3 Overall list
4 Some worth paying attention to the academic industry






Belief Network & Hopfield Network & Boltzman Machine & RBM structure at a glance 1 belief Network2 Hopfield Network3 Boltzman Machine 4 RBM Limited Boltzmann machine




RBM 1 Models


Using the formula, you can get

2 Solving the CD algorithm








DBN 1 Model 2 training oriented feature extraction oriented classification





DBM Model


CNN 1 Models
2 Training



Reference Documents

Artificial neural network deep learning MLP RBF RBM DBN DBM CNN Finishing Learning