Few-Shot Transfer Learning for Trust Score Calculation in Industrial Iot Devices

The Industrial Internet of Things (IIoT) is revolutionizing industries by enabling seamless connectivity and automation. However, the dynamic and distributed nature of IIoT networks poses significant challenges in ensuring the trustworthiness of devices, especially during onboarding with limited data. Traditional trust management methods often require extensive historical data, making them unsuitable for newly added devices with minimal information. To address this challenge, we propose a novel fewshot transfer learning approach for trust score calculation in IIoT environments. Our method leverages pre-training on probabilistic trust scores generated from a Markov Chain model, followed by fine-tuning with few-shot learning on real-world data. We evaluate the approach using three machine learning models: Random Forest, XGBoost, and Linear Regression, demonstrating that Random Forest outperforms the others in the few-shot scenario. We also analyze the computational complexity and energy consumption of the models to provide insights into the trade-offs between predictive performance, resource usage, and environmental impact. This work contributes a new perspective on trust management in IIoT, paving the way for more adaptive and data-efficient solutions.