Novel Feature Extraction Methods for Authentication via Mouse Dynamics with Semi-Supervised Learning

Using Machine Learning for Dynamic Authentication in Telehealth: A Tutorial

“…In some problems, such as authentication, there is no access to enough labeled data to train a supervised model; therefore, semi-supervised algorithms can address this issue. Yildirim et al [78] used a semi-supervised method of learning from behavioral biometrics of mouse dynamics data for authentication. This behavioral authentication model is very applicable to authenticate legal operators of telehealth platforms while connecting with patients or accessing their data.…”

“…Moreover, Shahbazi et al [94] used DRL in the context of blockchain to authenticate IoT devices in smart environment gateway, such as smart healthcare. Clustering [71,72,73,74] Behavioral Biometrics [78] Classification [80,81,82] k-NN [4,57] PCA [75,76] Outlier Detection [79] Feature Selection [38] RF [4,58,59,60,61,62,63,64] SVD [77] -Adaptive Authentication [80] SVM [4,65,66,67,68] --CAN [81,82]…”