Explainability through uncertainty: Trustworthy decision-making with neural networks
This work addresses the need for more robust and actionable machine learning systems in operations research, particularly for applications like educational data mining, though it is incremental by linking existing uncertainty methods to explainability.
The paper tackles the problem of neural networks being overconfident under distribution shifts by proposing a framework that uses uncertainty estimation as an explainable AI technique, resulting in improved trustworthiness and reduced misclassifications through human-in-the-loop rejection for uncertain observations.
Uncertainty is a key feature of any machine learning model and is particularly important in neural networks, which tend to be overconfident. This overconfidence is worrying under distribution shifts, where the model performance silently degrades as the data distribution diverges from the training data distribution. Uncertainty estimation offers a solution to overconfident models, communicating when the output should (not) be trusted. Although methods for uncertainty estimation have been developed, they have not been explicitly linked to the field of explainable artificial intelligence (XAI). Furthermore, literature in operations research ignores the actionability component of uncertainty estimation and does not consider distribution shifts. This work proposes a general uncertainty framework, with contributions being threefold: (i) uncertainty estimation in ML models is positioned as an XAI technique, giving local and model-specific explanations; (ii) classification with rejection is used to reduce misclassifications by bringing a human expert in the loop for uncertain observations; (iii) the framework is applied to a case study on neural networks in educational data mining subject to distribution shifts. Uncertainty as XAI improves the model's trustworthiness in downstream decision-making tasks, giving rise to more actionable and robust machine learning systems in operations research.