Stacking ensemble learning with heterogeneous models and selected feature subset for prediction of service trust in internet of medical things

Abstract

Abstract Recently, with the fast development of IoT, Internet of medical things (IoMT) has drawn wide attention from both industry and academia. However, pressing challenges exist in practical implementation of IoMT, such as service provision with stringent latency. To address the challenges, fog computing is generally employed in IoMT systems. However, it raises additional concerns of trust and security. To tackle the issue, the authors introduce the security measure of trust into this work, and a superior heterogeneous stacking ensemble learning measure for trustworthiness prediction (SEM‐TP) of fog services is proposed. Besides, to reduce unnecessary time cost incurred by unimportant features, an efficient voting‐based feature selection (FS) strategy called voting‐based feature selection method is proposed to select significant features, which is based on diverse FS measures. Extensive experiments are conducted and the results show that the proposed framework outperforms commonly used single classifiers and competing stacking models in terms of Accuracy, Precision, Recall, F1‐score, Kappa coefficient, and Hamming distance under different conditions, validating the effectiveness, robustness, and superiority of the proposed trustworthiness prediction and FS methods.

Keywords

• ensemble learning
• feature selection
• IoMT
• trustworthiness prediction