Abstract:
The launch of new applications and services has resulted in an increase in internet traffic in recent years. Consequently, network traffic management has become increasingly difficult. Various methods were suggested for classifying network traffic in order to achieve this task. The state of the art machine learning and deep learning models has been deployed by various researchers for proposed task. Boosting algorithms can also be used for proposed job. Boosting algorithms uses the concept of decision tree. They need little training time; a strong device is not required for model training. Therefore, the proposed study is conducted to classify network traffic using boosting algorithms such Extreme gradient boosting model (XGBM), Light gradient boosting model (LGBM), cat boost and ada boost and to compare the results of these models in terms of confusion matrix, accuracy, precision, recall, and FMeasure. Dataset used in the proposed study is online openly available named as VPN-NONVPN. Python and its libraries like tensor flow, keras, matplotlib and scikit learn etc. are used for simulation purposes. After completing simulation, it has been achieved that the XGBM achieved 90.41% of accuracy, precision as 96.39%, recall as 89.72% and f-measures as 92.91%, LGBM achieved 89.02% of accuracy, precision as 90.04%, and recall as 89.8% and f-measures as 89.83%. Cat boost achieved 86.87% of accuracy, recall as 83.97%, and precision as 89.43 and f-measure as 86.61%. After that ada boost achieved the accuracy of 83.07%, recall rate as 80%, precision as 85.25 and f-measures as 82.58%.
Page(s):
1-1
DOI:
DOI not available
Published:
Journal: Second International Conference on Computing Technologies, Tools and Applications (ICTAPP-24), June 4-6,2024 (Abstract Book), Volume: 0, Issue: 0, Year: 2024
Keywords:
Classification
,
LGBM
,
Network
,
Confusion Matrix
,
XGBM