Fuzzy-Rough-Set-Based Active Learning

摘要: Determining the informativeness of unlabeled samples is a key issue in active learning. One solution to this using sample's in- consistency between conditional features and decision labels. In paper, fuzzy-rough-set-based learning model proposed tackle problem. First, consistence degree labeled sample computed by lower approximations fuzzy rough set, which reflects its minimum membership class. Then, concept covering measure relationship un- samples. Afterward, memberships an be- longing different classes are based on degrees it. Finally, these used form selection criterion inconsistency. By applying Gaussian kernel-based similarity relation aforemen- tioned processes, support vector machine (SVM)-based scheme developed. Experimental results demonstrate effectiveness model.