Interpretable Fuzzy System for Malicious Domain Classification Using Projection Neural Network
Rajan Prasad, Praveen Kumar Shukla · International Journal of Wireless and Microwave Technologies · 2023
In this study, we suggest an interpretable fuzzy system for the classification of malicious domains.The proposed system is integration of Sugeno type fuzzy system and projection neural network, the main advantage of interpretable fuzzy system is to classify the patterns and self-explainable capability.Whereas the projection network is used to exact mapped fuzzy inference rules to the network's projection layer.On the other hands, the system is able to deal with large amount of real-time data.The proposed model is tested malicious URL datasets collected from Alexa.The experimental results show that the system is able to classify malicious domain with high accuracy and interpretability as compared to existing methods.The proposed model is usefull to classify malicious attacks and explain the couses behind the decision.The evaluation of model based on confusion matrices, ROC and the nauck index is used for the interpretability assessments.