CLASSIFICATION AND PREDICTION - Universität Wien

(19)
?X?
to the previous layer, 2
Peter Brezany Institut für Softwarewissenschaft, WS 2002 12
Backpropagation Algorithm
Algorithm: Backpropagation. Neural network learning for classification,
using the backpropagation algorithm.
Input: The training samples, samples; the learning rate, ;
a multilayer feed-forward network, network.
Slide 23
Output: A neural network trained to classify the samples.
Method:
(1) Initialize all weights and biases in network;
(2) while terminating condition is not satisfied
(3) for each training sample in samples
(4) // Propagate the inputs forward:
!
"$#&%('*),+ ) #.- )/10
(5) for each hidden or output layer unit
(6) ; //compute the net input of unit with respect
-3#4% 5
(7)
#>=?
// compute the output of each unit
57698;:9<
Backpropagation Algorithm (2)
(8) // Backpropagate the errors:
(9) for each unit in the output layer
(10) @!ABA#&%(-3#§CEDGFH-I#.JKCMLN#!FO-3#J ; // compute the error
(11) for each unit in the hidden layers, from the last to the first hidden layer
Slide 24
(12) @!ABA#&%(-3#§CEDGFH-I#.J '*P @QABA P + # P ; //compute the error with respect to
the next higher layer, R
+ ) #
) S&+ ) #Q%TCMJE@!ABA#U-
(13) for each weight in network
(14) ; // weight increment
(15) + ) #Q%V+ ) # / S&+ ) # ;
?
// weight update
0 #
S #&%TCMJW@QABA# 0
(16) for each bias in network
(17) ; // bias increment
(18) 0 #4% 0 # / S 0 # ;
?
// bias update