Biuro KarierPolitechnika Warszawska

Topic: “Sponsored content in contextual bandits. Combining interventional and observational data in sequential decision making.”

About the topic: Contextual multi-armed bandit is a prevalent framework for modelling the sequential allocation of resources under uncertainty. Among abundance of applications of the model are recommender and advertising systems as well as clinical trials. Notably, these systems often feature sponsored content through mechanisms like lobbying or marketing. In this work, we introduce the Authoritarian Sponsor model - a novel approach to modelling sponsored content within the bandit framework. Among the consequences of existence of sponsoring are forced exploration, unknown propensity score and possibly a confounded treatment assignment mechanism.

Presenter: Hubert Drazkowski, finalist of the 5th Annual BNY Mellon Science​ Master’s Thesis Competition in Applied Mathematics

Special Guest:
Upcoming Science Seminar will also host our special guest: Katie Hysenbegasi, Head of Risk Modelling and Analytics at BNY Mellon, who will share her insight on Risk Modelling and Analytics and BNY Mellon as the one of the largest custodian bank.