Zum Inhalt springenZur Suche springen


Petr Spelda & Vit Stritecky: "Sociotechnically Alligned & Robust Machine Learning" (Online)

Philosophie Philosophie II Forschungsseminar

Institut für Philosophie

Seniorprofessur für Theoretische Philosophie

Im Rahmen des Forschungskolloquiums laden wir ein zum Vortrag von

Dr. Petr Spelda (Charles University, Prague)
Dienstag, 06.12.2022 18:30 – 20:00 Uhr (Online)
Sociotechnically Aligned & Robust Machine Learning

Abstract: The talk will introduce issues associated with managing high-risk machine learning systems for protecting people’s safety, security, and fundamental rights. Key stages of high-risk machine learning lifecycles will be explained and related to different kinds of security analysis that can help to reduce the risk surface of high-risk machine learning systems. The concept of sociotechnical robustness, integrating social and technical requirements on inferential properties of machine learning systems, will be introduced and its social choice and no-regret aspects explained. Among the motivations of the talk is to provide a sound basis for discussing the viability of different types of legal regulatory instruments targeting high-risk machine learning.

Speaker: Petr Spelda is an Assistant Professor at the Department of Security Studies, Charles University, Prague. His research focuses on safety and security of machine learning and on treating robustness of machine learning systems as a sociotechnical problem. His works on the topic appeared in ACM Computing Surveys, Futures, or Axiomathes. How to join the talk: The talk will be held online via Zoom. In order to receive the Zoom link for joining the talk please write to maria.sekatskaya@hhu.de.



06.12.2022, 18:30 Uhr - 20:00 Uhr