This presentation explores an innovative algorithm for conducting Bayesian inference, bypassing the need for employing the likelihood. The method will be thoroughly explained, accompanied by real-world illustrations of its application within the domains of insurance and finance. Additionally, we will direct our attention towards a specific use case: parameter recovery of a random sum, commonly referred to as decompounding.

 

Speaker Biography: Pierre-O Goffard is an associate professor at l'Institut de science financière et d’assurances, a graduate school specializing in actuarial science and part of Universite Lyon 1. His research interests focus on probabilities and statistics applied to risk management of insurance companies and optimization of blockchain systems.