Adaptive Bi-Recommendation and Self-Improving Network for Heterogeneous Domain Adaptation-Assisted IoT Intrusion Detection

As Internet of Things (IoT) devices become prevalent, using intrusion detection to protect IoT from malicious intrusions is vital importance. However, the data scarcity hinders effectiveness traditional methods. To tackle this issue, in article, we propose adaptive bi-recommendation and self-improving network (ABRSI) based on unsupervised heterogeneous domain adaptation (HDA). The ABRSI transfers enrich knowledge a data-rich source facilitate effective for data-scarce target domains. achieves fine-grained transfer via matching. Matching interests two recommender systems (RSs) alignment categories shared feature space form mutual-benefit loop. Besides, uses mechanism, autonomously improving four ways. A hard pseudo label (PL) voting mechanism jointly considers RS decision relationship information promote more accurate PL assignment. diversity participation during transfer, instances failing be assigned with will probabilistic soft PL, forming hybrid pseudo-labeling strategy. Meanwhile, also makes pseudo-labels globally diverse individually certain. Finally, an error learning utilized adversarially exploit factors that causes ambiguity learns through both current previous knowledge, preventing forgetfulness. Holistically, these mechanisms model boosts accuracy HDA-assisted transfer. Comprehensive experiments several sets demonstrate state-of-the-art performance method, outperforming its counterparts by 9.2%, verify constituting components ABRSI’s overall efficiency.