.png){kind=icon}
返回
An active learning framework using deep Q-network for zero-day attack detection
.png){kind=icon}
PRE
AI
DOI：10.1016/j.cose.2024.103713.png)
.png){kind=icon}
原文求助.png){kind=icon}
分享.png){kind=icon}
收藏
摘要
En 中文
Zero-day attacks from highly dynamic and complex cyberspace make severe threats to the existing network intrusion detection systems. Robust and intelligent solutions are required to handle the rapidly evolving threats from internet environment. In this paper, a novel active learning framework based on Deep Q-Network (DQN) is proposed for zero-day attacks detection. The framework is composed of network intrusion detection systems classifier, sample selection strategy and annotator. Deep Q-Network model serves as an intelligent control component to select the zero-day samples for labeling with the probability distribution. Moreover, the Bidirectional Long Short-Term Memory (BiLSTM) network is integrated into DQN model to form the selecting policy to analyze the temporal correlation within the static classification context. Meanwhile, Euclidean distance function is adopted for labeling the selected samples to achieve an accurate labeling result. Two famous datasets named NSL_KDD and UNSW_NB15 in intrusion detection field are tested for evaluate the performance of our proposed framework. Compared to other machine learning methods, the approach proposed in this paper demonstrates better universality and generalizability. This indicates that the method is more capable of reliably identifying and detecting network intrusion behaviors, which holds significant importance for further security research and practical applications.
Keyword:
Network intrusion detection systems
Active learning
Deep Q -network
Annotator
期刊
C
IF:
5.4
论文数:
4.6K
被引数:
1.4W
机构
暂无机构信息
引用论文
Anomaly-based intrusion detection system for IoT networks through deep learning model基于深度学习模型的物联网网络异常入侵检测系统
OAAI
OAAI分享收藏Study of bi-directional buck-boost converter topologies for application in electrical vehicle motor drives应用于电动汽车电机驱动的双向buck-boost变换器拓扑研究
PREAI
分享收藏Deep Belief Network Integrating Improved Kernel-Based Extreme Learning Machine for Network Intrusion Detection.png)
IEEE ACCESSOAAI
IEEE ACCESSIF3.6
OAAI分享收藏Adventures in data analysis: a systematic review of Deep Learning techniques for pattern recognition in cyber-physical-social systems数据分析的冒险: 网络物理社会系统中模式识别的深度学习技术的系统回顾
PREAI
分享收藏
查看更多内容