LLM graph hallucinations LLMs hallucinate graphs too! (here prompted for the Karate Club Graph)


I am a researcher at Inria, Brittany, France (ARTISHAU team). I am leading the scientific council of the Société Informatique de France, and I am the president of an association promoting free software for mail/hosting in Brittany.

My current research is studying black-box algorithms (or links to papers) in the context of of decision-making algorithms (neural-network models, LLMs, etc). In other words, I am interested in the audit/understanding of algorithms from the user-side (FR). I co-organize the WAAA workshop on algorithmic audits of algorithms, in an attempt to link several research domains on this societal question. e.g.:

Selected publications

  • Setting the Record Straighter on Shadow Banning, Erwan Le Merrer, Benoît Morgan, Gilles Trédan. In INFOCOM, 2021. PDF
  • Remote Explainability faces the bouncer problem, Erwan Le Merrer, Gilles Trédan. In Nature Machine Intelligence, 2020. PDF
  • Adversarial Frontier Stitching for Remote Neural Network Watermarking, Erwan Le Merrer, Patrick Perez, Gilles Trédan. In Neural Comput & Applic 32, 9233–9244, 2020. PDF
  • Second order centrality: Distributed assessment of nodes criticity in complex networks, AM Kermarrec, E Le Merrer, B Sericola, G Trédan. Computer Communications 34 (5), 619-628, 2011. PDF & code in NetworkX

Last PCs

  • KDD’24: 30TH ACM SIGKDD conference on knowledge discovery and data mining
  • ECAI’24: 27TH European conference on artificial intelligence
  • AIsec’24: 17TH ACM Workshop on Artificial Intelligence and Security
  • ECML/PKDD’24: European Conference on Machine Learning and Data Mining
  • SDM’24: SIAM International Conference on Data Mining
  • KDD’23: 29TH ACM SIGKDD conference on knowledge discovery and data mining