Comparison of Supervised Learning and Reinforcement Learning in Intrusion Domain

Sengupta, Nandita; Sil, Jaya

doi:10.1007/978-3-642-31686-9_63

Cited by 2 publications

(1 citation statement)

References 5 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Intelligent intrusion detection has been an area of research for quite some time; many approaches have been developed using reinforcement learning [8,9]. Some of these approaches include providing a response to a network attack.…”

Section: List Of Tablesmentioning

confidence: 99%

Intelligent Threat-Aware Response System in Software-Defined Networks

Goswami

View full text Add to dashboard Cite

INTELLIGENT THREAT-AWARE RESPONSE SYSTEM IN SOFTWARE-DEFINED NETWORKS by Kunal Goswami Software-defined networks decouple the control plane from the data plane, enabling researchers to evaluate protocols and network configurations through the centralized point of control, the controller. They provide easy management and automation, scalability, and flexibility in the traditional computer network. In spite of these advantages, software-defined networks fall prey to various denial-of-service attacks specific to network protocols and applications despite their simplicity. There is a need to implement intelligence in the controller as a countermeasure for not only the various types of denial-of-service attacks but also the increasing sophistication involved in them. In this paper, an intelligent threat-aware response system is proposed for defending against any attack by using reinforcement learning. Reinforcement learning can acquire intelligence for detection and reactive actions through experience with various attacks. This experience is obtained from interactions with the computer network through the controller. With the combination of reinforcement learning and the software-defined networking controller, the goal of the autonomous threat response system can be achieved. ACKNOWLEDGMENTS I would like to thank Dr. Younghee Park for her continuous guidance and support throughout my time as a graduate student. I am greatly indebted to Dr. Park for all the technical concepts and research styles that I learned from her. Another person who had an impact on my work is Dr. Chungsik Song. I would like to thank him for his valuable insights and comments on the algorithmic approach for this work. I offer my gratitude to the committee members for their expert advice on how to formulate the thesis, evaluation, and presentation. v

show abstract

Section: List Of Tablesmentioning

confidence: 99%