Federated learning meets explainable AI at the edge of things
2024
Federated learning (FL) and explainable artificial intelligence (XAI) stand as pivotal advancements in the realm of modern Internet of Things (IoT) applications, offering innovative paradigms to address critical challenges in decentralized learning and model interpretability. FL, a paradigmatic shift in machine learning (ML), has gained prominence due to its efficacy in training models across a network of edge devices while preserving data privacy. Unlike traditional centralized approaches, FL operates by distributing the learning process among a multitude of local devices, enabling model training on data generated at the edge without the need for raw data aggregation in a central repository. This decentralized learning framework not only safeguards sensitive user information but also empowers IoT systems by harnessing the collective intelligence of distributed devices, thereby enhancing model performance and adaptability in dynamic edge environments. On the other hand, XAI has emerged as a critical attribute in contemporary IoT ecosystems, aiming to demystify the decision-making process of complex ML models. In essence, XAI endeavors to elucidate the reasoning behind AI-driven decisions, enabling stakeholders to comprehend and trust model outputs. With the surge of AI-powered applications in IoT, interpretability becomes paramount, particularly in critical domains where transparent and accountable decision-making mechanisms are essential [1].