Publications & Papers

Optimal Transport Prototypes and Symbolic Logic Combination for Self-Explainable Graph Neural Networks

Elouan Vincent, Julien Perez, Marc Plantevit, Céline Robardet

European Conference on Machine Learning and Principles and Practice of Knowledge Discovery in Databases (ECML PKDD), Naples, Italy, September 2026

We introduce GELATO (Graph Explainability via Logic And Transport Optimization), a self-explainable graph classifier that integrates optimal transport, prototype grounding, and symbolic logic to produce faithful and human-readable explanations grounded in real graph structures.

Diffusion for Explainable Unsupervised Anomaly Detection

Elouan Vincent, Alexandre Dréan, Julien Perez, Marc Plantevit, Céline Robardet

IEEE International Conference on Data Science and Advanced Analytics (DSAA), 2025

This work explores the use of diffusion processes for enhancing the interpretability of unsupervised anomaly detection methods, combining the power of diffusion models with explainable AI techniques.

Improving the Quality of Rule-Based GNN Explanations

Ataollah Kamal, Elouan Vincent, Marc Plantevit, Céline Robardet

Workshop on eXplainable Knowledge Discovery in Data Mining (XKDD), ECML PKDD 2022, Grenoble, France, September 19-23, 2022, pp.467–482

This work focuses on enhancing the quality and interpretability of rule-based explanations for Graph Neural Networks, addressing the challenge of making GNN decisions more transparent and understandable through improved rule extraction methods.