Manolis at ICAIF24 in New York, USA.
10 December 2024
Our Ph.D. student, Manolis Panagiotou, attended the ACM International Conference on AI in Finance (ICAIF24) in Brooklyn, New York, and delivered an oral presentation of our paper titled "TABCF: Counterfactual Explanations for Tabular Data Using a Transformer-Based VAE", co-authored by Prof. Ntoutsi.
In this work, we generate counterfactual explanations for AI decisions in critical areas like finance. In such real-world applications, data often comes in tabular form, presenting specific challenges like mixed feature types, inherent imbalances, and complex feature interdependencies. We address issues with current methods related to data handling, such as feature-type bias, by leveraging a transformer-based VAE specifically tailored for tabular data.
Please follow the link for the full paper.