A New Feature Scoring Method in Keystroke Dynamics-Based User Authentications

Abstract

In recent years, computing devices have become widely distributed, and the accumulated data from these devices are growing rapidly, especially as they are increasingly equipped with various sensors and RF communication capabilities. Data science, including machine learning technology, has contributed to the better handling the large amounts of data and feature selection techniques have been a useful strategy. As data amounts continue to grow, scaling features will became crucial in data science. In this paper, we propose a novel filter-based feature-selection method in the context of keystroke dynamics authentication. In particular, we propose a new feature-scoring method and apply it to keystroke-dynamics-based authentications. We implement keystroke-dynamics-based authentications multi-factored with PIN-based authentications and collect data from actual users' testing experiments. Then, we apply our feature-selection method and compare the performance with that when using all of the features and existing feature-selection methods. Our experimental results show that the classification performance by the proposed method is superior to those of the other methods by up to 21.8%. Moreover, our method provides security to other users' data sets, as the method utilizes only mean values from imposter data. Our feature-selection method contributes to improving the quality of keystroke dynamics authentications without user privacy issues. More generally, our method can also be applied to other data-mining data sets, such as IoT sensor data sets.

Keystroke Dynamics-Based Authentication Using Unique Keypad

“…In the present study, the same feature selection method used in [ 4 ] was employed to improve classification performance, based on the extracted data. More specifically, the data features were filtered according to the calculated feature ranking: first, those in the lower 10% group were filtered, and then those in the lower 20% group were filtered.…”

“…Thus, it is necessary to select some features that may be helpful to improve the ability of the system to classify normal and abnormal users who attempt authentication. In the present study, the feature scoring method used in [ 4 ] was selected as a feature selection method. The 152 features collected in the present study were ranked according to the values obtained using the feature scoring method proposed in [ 4 ].…”

“…In the present study, the feature scoring method used in [ 4 ] was selected as a feature selection method. The 152 features collected in the present study were ranked according to the values obtained using the feature scoring method proposed in [ 4 ]. The feature scoring equation in [ 4 ] can be expressed as Equation ( 9 ), which corresponds to the difference between the variation coefficient of the abnormal user data ( ) and the variation coefficient of the normal user data ( ) divided by the variation coefficient of the normal user data ( ).…”

“…The results showed that the user classification performance was the best when the most frequently typed n-graph selection method was applied. In a 2020 study, based on the trimmed mean and variation coefficients of the collected data, feature scores were determined and compared [ 4 ]. Features with low scores were discarded to improve classification performance.…”

“…Unlike existing methods, the method proposed in the present study was designed to improve user classification performance by assigning a unique keypad to the normal user during the user classification process. Using the feature selection algorithm presented in [ 4 ], the data collected from the unique keypad was subject to feature selection and then preprocessing, followed by user classification tests using three types of classifiers: Linear support vector machine (SVM), One-Class SVM, and Manhattan distance-based classifiers. The test results obtained using the unique keypad were then compared with those obtained using a conventional keypad.…”