##plugins.themes.bootstrap3.article.main##

Poonam Lohiya

Gajendra Bamnote

Abstract

Protocol prediction in network traffic allows for smarter, more effective network management by anticipating the types of data that will flow through the network and regulating resources consequently.The prominence is on using Optuna, a framework for fine-tuning model configuration parameters that maximize model performance. A deep learning-based model that was tuned for tasks involving the classification of network traffic is created using TensorFlow/Keras. In order to increase classification accuracy, a variety of model setup parameters by employing Optuna's effective search methods and pruning processes is implemented. This AI-based network traffic classification model attained high performance, with 90.48% accuracy, 90.70% precision, 90.47% recall, and a 90.52% F1-score. The final model showed that optimizing model configuration parameters improve network traffic prediction and categorization, as seen by the distinguished gains in accuracy and other evaluation metrics that it attained.

##plugins.themes.bootstrap3.article.details##