Figure 2
From: Discriminant non-stationary signal features’ clustering using hard and fuzzy cluster labeling
The block diagram of our proposed method for discriminant feature clustering. In Stage 1, a set of clusters is identified using an unsupervised clustering method which is simultaneously applied to the training feature vectors extracted from the different classes. In Stage 2, based on the distribution of the train signals in each cluster, it will be decided whether a cluster is normal or abnormal. In the test stage, the class label of each signal is assigned based on the membership of the relying feature vectors to one of these clusters.