Best Poster at ICDM 2025
We are proud to announce that the research paper “Explanations Go Linear: Post-hoc Explainability for Tabular Data with Interpretable Meta-Encoding” (Piaggesi et al., 2025) has been awarded Best Poster at the 2025 IEEE International Conference on Data Mining (ICDM).
The work, authored by Simone Piaggesi, Riccardo Guidotti, Fosca Giannotti, and Dino Pedreschi, introduces ILLUME, a novel methodology designed to enhance the transparency of black-box models on tabular data.
About the Work
The paper presents a breakthrough in post-hoc explainability. By leveraging an interpretable meta-encoding, the proposed approach maps complex data representations into a linear, understandable space. This technique allows for the generation of explanations that are not only accurate but also intrinsically interpretable, bridging the gap between complex machine learning models and human understanding.
This recognition at ICDM 2025, one of the premier conferences in the field of data mining, underscores the significance and impact of this research within the global scientific community.