Inference Techniques for Resilience

Grunddaten zum Vortrag

Informationen zur Veranstaltung

Zusammenfassung

According to Avizienis et al. (2004), resilience is a system’s ability to remain operational – although at potentially lower operational levels – when exposed to stressors and to adapt its functioning if those stressors persist. When working with formal models in the background, probabilistic inference may be used to predict or identify stressors, or compute necessary adaptations to formal models for a better representation of the system under duress. This leads to an algorithmic technical side to resilience. The inference part should keep going even if there is a sudden influx in observations or queries, or it resume as fast as possible if indeed a change in the model has occurred. This talk looks at resilience for inference. Specifically, it highlights how methods from statistical relational AI can help build more resilient algorithms using its inherent characteristics to support a system's overall resilience.

Vortragende der Universität Münster

Braun, Tanya

Juniorprofessur für Praktische Informatik - Moderne Aspekte der Verarbeitung von Daten / Data Science (Prof. Braun)