An explainable big transfer learning approach for IoT-based safety management in smart factories

Abstract

The integration of the Internet of Things (IoT) in smart factories enhances management through real-time monitoring and data analytics, while Artificial Intelligence (AI) automates processes and boosts efficiency. However, AI systems require vast amounts of data and substantial training time, facing challenges such as domain discrepancies, limited labeled data, negative transfer, sample selection bias, and computational complexity. Additionally, the opaque nature of AI models raises transparency issues, making it difficult for human operators to trust and interpret AI decisions. To address these challenges, this paper proposes an IoT-based safety management scheme for smart factories, utilizing advanced technologies to enhance safety and operational efficiency. The proposed approach integrates robust deep learning (DL) models developed through big transfer learning (BiTL) and is augmented with explainable AI (XAI) to ensure transparency and reliability in safety management. The major contributions of this work include designing a comprehensive IoT-based safety framework, conducting a detailed case study to optimize DL model performance using BiTL, and establishing an experimental environment for thorough validation. The findings demonstrate that the proposed system not only meets but also exceeds the performance of existing safety management solutions, offering a transparent, trustworthy, and highly effective AI-driven safety management system for modern smart factories.