Iot Profiling With Transductive Transfer Learning

Abstract

The rapid proliferation of Internet of Things (IoT) devices across various domains has introduced significant challenges in device management and security profiling. Traditional machine learning approaches to IoT profiling often rely on large amounts of labeled data, which are difficult to obtain in dynamic and heterogeneous IoT environments. This paper proposes a novel framework for IoT device profiling using Transductive Transfer Learning (TTL), a technique that enables knowledge transfer from a labeled source domain to an unlabeled target domain. The proposed system effectively classifies IoT devices and assesses their vulnerabilities by leveraging behavioral features extracted from network traffic data.
The methodology incorporates statistical feature selection techniques and evaluates the performance of multiple machine learning models, including Random Forest, Gradient Boosting, and Support Vector Machines. To validate the transferability and robustness of the approach, extensive experiments were conducted using diverse datasets such as CIC IoT 2022, IMC 2019, and IoT Sentinel. The results demonstrate high classification accuracy and reliable vulnerability assessment across varying environments. This work contributes to advancing secure and scalable IoT network management by reducing the dependency on labeled data and enabling real-time device identification and risk evaluation.