Utilizing a machine learning algorithm to choose a significant traffic identification system

Abstract

The tremendous advancements in computing, sensing, and cognitive-based revolution have made way for critical infrastructure needed to improve the internet for more extensive applications. However, the computation performed by devices and the ability to communicate is limited due to the use of numerous devices for storage. Among the most complex model scenarios, a framework based on a Traffic Identification System (TIS) is a crucial answer. The Web-based brute force and botnet attacks are linked in this initial phase. To forecast the samples using the adaptive bootstrap method, the second phase entails creating a classifier on the training dataset. The normalization model and features are combined in the model, which then identifies traffic attacks on the network. Based on voting mechanisms from several decision trees utilized in the mode, an ensemble model using the Adaptive Boosting Classifier and Random Forest Algorithm (ABC-RFA) method is employed to construct the classification model. According to the overhead analysis of the dataset used for testing, the suggested model's overall performance is approximately 98.90 %, much better than the conventional models currently in use. Overall Precision, Specificity, Rate of Detection, And Accuracy Were All Attained.

Keywords

• Adaptive bootstrap
• Classification
• Random Forest (RF)
• Ensemble Model