On using wearable devices to steal your passwords: A fuzzy inference approach

Abstract

The security of wearable devices user’s privacy data has become more and more concerned because of the high accuracy of the embedded sensors. Existing methods of obtaining privacy data often rely on installations of dedicated hardware, or accurate numerical calculation of sensor data, which do not have flexible adaptability. In this paper we utilize a multi-SVM and a KNN classifier using only accelerometer data and fuzzy coordinates to get the privacy data such as password directly with a higher accuracy.