.png){kind=icon}
.png){kind=icon}
Return
M2VT-IDS: A multi-task multi-view learning architecture for designing IoT intrusion detection system
.png){kind=icon}
PRE
AI
DOI：10.1016/j.iot.2024.101102.png)
.png){kind=icon}
Original request for help.png){kind=icon}
Share.png){kind=icon}
Save
Abstract
En 中文
With the rapidly growing frequency of security incidents in the Internet of Things (IoT), intrusion detection systems (IDS) have gained increasing attention in recent years. They have been applied in a variety of tasks, such as anomaly detection, attack identification, and device identification. The intrusion detection approaches based on deep learning have achieved promising performance owing to their capability of automatically discovering generalizable patterns. Nevertheless, the network traffic representations fed into neural networks are often designed for specific tasks, and their efficiency significantly depends on the knowledge of experts. In this paper, using a multi -view representation of network traffic with a multi -task learning architecture, we design a multi -task multi -view IoT intrusion detection system (M2VTIDS) that can provide multiple intrusion detection capabilities with high detection accuracy. We first construct a packet -wise representation of IoT traffic from a multi -view perspective, including a spatio-temporal series view, a header field pattern view, and a payload semantic view. Then, we design a two -stage multi -task learning architecture that consists of a multi -view shared network and a task -specific attention network, which can simultaneously realize anomaly detection, attack identification, and device identification tasks. Experimental results based on three popular IoT traffic datasets show that the proposed M2VT-IDS can achieve higher accuracy in multiple intrusion detection tasks when compared with other state-of-the-art specialized IDS schemes.
Keywords:
Internet of Things
Intrusion detection system
Multi-view representation
Multi-task learning
Journal
IF:
7.6
Papers:
1.9K
Citations:
6.9K
Organization
No organization information available