Skip to main content
. 2017 Jun 13;17(6):1370. doi: 10.3390/s17061370

Figure 4.

Figure 4

The learning structure of PAC. Different from ordinary pattern recognition method, PAC chooses RPs before training LS-SVM during the initialing process, and adds an updating process after labeling unlabeled samples. The updating process is composed of attractive zone decision (to select unlabeled samples with high confidence), updating RPs with selected samples, and universal incremental LS-SVM with updated RPs.