A Cycle Deep Belief Network Model for Multivariate Time Series Classification

Algorithm 1

DBN train algorithm.

Begin

Initialize (the learning rate), (the number of

DBN layers), (the number of hidden unites in layer

and ), epochs, DBN().parameters

(the parameters of each layer and ).

 Input data ;

 ;

 ;

 do

 trainRBM(1, DBN().parameters);

 ++;

 compute according to Equation (3)

 ;

 while ()

 End.