Seminar: Making explainable artificial intelligence understandable

Date and time 27. 10. 2026 15:00 - 16:30
Room NB 473 and online: https://teams.microsoft.com/meet/3656063550482009?p=zdF5NFdCRxeVnkEjl2

Szymon Bobek

Speakers: GEIST Research Group, Jagiellonian University

What the talk will cover:

  • Lessons from industrial use-cases: predictive maintenance,explainable anomaly detection, and visual quality inspection.
  • Evidence from  a user-centric evaluation of explanation methods.
  • Explainability methods based on rules, prototypes, counterfactuals, and feature attribution.
  • Open research directions in explainability with neurosymbolic AI, LLMs, VLMs, eye-tracking analysis and conversational systems for interactive, multimodal and expert-aligned explanations.