Explainable Artificial Intelligence for Trustworthy Internet of Things

A major challenge for machine learning solutions is that their efficiency in real-world applications is constrained by the current lack of ability of the machine to explain its decisions and activities to human users. Biases based on race, gender, age or location have been a long-standing risk in training AI models. Furthermore, AI model performance can degrade because production data differs from training data.