Intellectual Outlier Identification in Network Anamoly Using Deep Reinforcement Learning

Intellectual Outlier Identification in Network Anamoly Using Deep Reinforcement Learning

Authors- Assistant Professor Kiruthika.S, Danushya.V, Jananisri.V, Jayashree.M, Juniper Zibiah.F.C

Abstract-The rapid expansion of Internet of Things (IoT) devices has intensified the challenge of securing network infrastructures from several attacks. Traditional security measures often struggle with the diverse and large-scale nature of IoT traffic. This paper proposes an advanced approach to detection and mitigation by integrating Deep Reinforcement Learning (DRL) with the VGG16 convolutional neural network. The system employs VGG16 to extract and analyze features from network traffic, identifying potential anomalies indicative of attacks. These features are then processed by a DRL agent, which learns and optimizes defensive strategies based on real-time feedback. This adaptive learning approach enhances detection accuracy and response effectiveness by continuously updating its strategies based on evolving attack patterns. The system’s dynamic nature allows it to scale with the growing complexity of IoT networks, offering a robust solution to improve network security. This paper details the integration of DRL and VGG16, presents experimental results demonstrating the system’s effectiveness, and discusses its potential to significantly advance IoT security management.

DOI: /10.61463/ijset.vol.12.issue6.372