Random Deep Belief Networks for Recognizing Emotions from Speech Signals

Algorithm 1

RDBN.

Input. Training speech signals , the ensemble size , and the input speech signal

Output. The emotion label

Training Stage

 () Extract the features for each speech signal in

     where is the feature vector of

 () Create random subspaces from

 () Create deep belief networks from

 () Create the base classifiers for the ensemble

Testing Stage

 () Extract the features for the speech signal

 () Create random subspaces from

 () Input each random subspace into

 () Take the output of each as the input of

 () Assign the emotion label for by the majority voting, where is the Boolean function