Rosina Weber, PhD
Explainable AI provides human users with tools to understand the output of machine learning algorithms. One of these tools, feature attributions, enables users to know the contribution of each feature of an AI model to that model’s predictions.
In a new article, upcoming in the Nov. 25, 2025 issue of AI journal Knowledge-Based Systems, CCI Professor of information science and computer science Rosina Weber, PhD and fellow researchers explain one challenge in developing feature attributions in regression tasks (or, supervised learning by an AI model to predict numerical values) is a lack of data sets to evaluate feature attributions.
Weber et al. demonstrate a new way to generate those data while meeting privacy requirements; further, the team proposes using additive case-based reasoning and the coefficient of quartile variation (CQV). All three methods will provide “consensus and benchmarking in XAI evaluation for regression models" that are currently lacking, the team stated in their article, “Privacy-preserving ground-truth data for evaluating additive feature attribution in regression models with additive CBR and CQV.”
“This is the result of the collaboration with Shahina Begum, Mobyen Uddin Ahmed and Mir Riyan Islam from Mälardalens University, Sweden,” said Weber, with the CCI professor collaborating as an international XAI expert to equip their applied systems with explainability.
The team’s real-world focus has centered around integrating explainability into a flight-delay prediction system from the Mälardalens team that collaborates with Eurocontrol, the European organization for the safety of air navigation.
Professor Weber continues to make significant contributions to the field of explainable AI, having delivered the impactful keynote “XAI Is in Trouble,” based on the article with the same name, at the prestigious XAI Workshop, part of the International Joint Conference on Artificial Intelligence (IJCAI) in August.
In February, Weber proudly joined the Executive Council of the Association for the Advancement of Artificial Intelligence (AAAI). As part of her work as a council member, she is a member of the Policy and the Conference committees.