A Dominant Feature Selection Method for Deep Learning Based Traffic Classification Using A Genetic Algorithm

Abstract

Internet data handling is becoming a challenging issue to networking organizations now a days. To have an efficient throughput of network functions various traffic classification techniques were propose so far. These types describe encrypted traffic classification which uses the support of deep learning approaches. While packet inspecting payload is another stuff of classification. A malfunction of the deep learning model may occur if the training dataset includes malicious or erroneous data. Security and confidentiality of users data while in networking is another open issue which can be solved by using Explainable artificial intelligence (XAI) somehow. In this paper, we propose a strategy for making sense of the functioning system of deep learning-based traffic grouping as a technique for XAI in view of a hereditary calculation. We depict the component of the deep learning-based traffic classification by measuring the significance of each element by using genetic algorithms. Moreover, we influence the hereditary calculation to produce an element determination veil that chooses significant highlights in the whole list of capabilities. To exhibit the proposed clarification technique, we carried out a deep-learning based traffic classifier with an exactness of roughly 96.55%. Likewise, we present the significance of each component got from the proposed clarification technique by characterizing the predominance rate.