In this paper we present a formal and extensible framework for a context-aware, and adaptive multi-factor authentication (CAA-MFA). CAA-MFA supports the selection of k-number of random factors based on the risk of the user’s login context. Each authentication factor in association with its source device has an associated trust score. We use these trust scores to evaluate individual factor’s feasibility for user authentication, and the collective feasibility of the k-number of factors. We use context-based reasoning (Protege) to derive the risk of the user login and therefore the minimum number of required factors and their trust scores. Our system will randomly select a set of factors that satisfies the login requirements. Next, we propose a continuous assessment method of the users login sessions. This serves as a second layer of defense against unauthorized user logins. We classify user logins into suspicious (unauthorized) and non-suspicious (authorized) categories. For this, we use a combination of Risk Level assessment (RLA) computation and known attack patterns. Our empirical results indicate that this integrated RLA guided approach improves resistance to spoofing and phishing attacks without compromising usability. We tested our systems performance using F1-score and CAA-MFA achieves a score of 0.985.

Behavioral biometrics is an emerging trend due to their cost-effectiveness and non-intrusive implementations that support remote access for user identification. This is the case especially in recent times of social distancing and working from home arrangements, where online attendance is the preferred option in contrast to physical presence. In this work, we explore the limitations of mouse dynamics authentication by impersonating legitimate user mouse action sequences. Specifically, towards that aim, we develop a novel generative WC-DCGAN model to generate highly accurate fake user action sequences. We apply our WC-DCGAN to this problem and show that it causes the target classifier can be tricked into identifying a fraudster as a legitimate user. WC-DCGAN has several benefits, including: achieving dominated convergence, hence implying the existence of solutions and optimal discriminator regardless of data and generator distributions; and acting as an unsupervised model for a fixed class label and generator. Experiments are conducted to verify these points. Subsequently, we analyzed the cause of misclassifications, and propose a novel mouse dynamics strategy that offers much tighter authentication with significant reductions in misclassification events.

Apriori-based rule generators, which are powered by the DIS-Apriori algorithm and the NIS-Apriori algorithm, are applied to analyze the data sets available in the IEEE BigData 2019 Cup: Suspicious Network Event Recognition. Then, each missing value in the test data set is decided by using the obtained rules. The advantage of our rule-based model is that the obtained rules are very easy to understand in comparison with other ”black-box” machine learning models. Furthermore, two algorithms preserve the logical property ”completeness,” so they generate rules without excess and deficiency. In evaluation, the AUC measure seems unfavorable to our model, so we employed 3-fold cross-validation for the training data set, and we obtained a 94% mean score. This result ensures the validity of our model. We report several meaningful results in this experiment, as well as the estimation of missing values.