Online/Incremental Learning to Mitigate Concept Drift in Network Traffic Classification
Abstract
Communication networks play a large role in our everyday lives. COVID-19 pandemic in 2020 highlighted their importance as most jobs had to be moved to remote work environments. It is possible that the spread of the virus, the death toll, and the economic consequences would have been much worse without communication networks. To remove sole dependence on one equipment vendor, networks are heterogeneous by design. Due to this, as well as their increasing size, network management has become overwhelming for network managers. For this reason, automating network management will have a significant positive impact. Machine learning and software defined networking facilitate automating network management. Classifying network traffic to its application of origin allows us to further contribute to this industry's vision. Most of the research behind NTC has been successful at classifying network traffic via traditional batch machine learning methods. However, network traffic suffers from a phenomenon known as concept drift which leads to a decline in classification performance. Online/incremental learning algorithms are designed to handle these environments and can assist in mitigating concept drift in network traffic. Our research compares traditional batch learning algorithms with online/incremental algorithms at handling concept drift in network traffic. Also presented is the concept of homeostatic learning, patterned after the biological mechanism known as homeostasis in living organisms that handles non stational environments.
Subject Area
Computer Engineering|Computer science|Educational technology
Recommended Citation
De La Rosa, Alberto, "Online/Incremental Learning to Mitigate Concept Drift in Network Traffic Classification" (2022). ETD Collection for University of Texas, El Paso. AAI30242042.
https://scholarworks.utep.edu/dissertations/AAI30242042