Home

Lecture Booking.com

in 2 weeks
Thursday 9 feb 17:00 - 18:00
C0.05

Free


Booking.com needs to make millions of decisions each day that optimize the experience of their guests.

However, learning how to make these decisions is notoriously difficult due to something called a Bandit problem. In Bandit problems, you need to learn what the optimal decision is, while you can only ever observe the outcome of the actions you took. For example, if you recommend someone to search for hotels in Amsterdam, you will not know for sure what would have happened if you had shown hotels in Berlin instead. Answering these counterfactual questions is at the heart of solving Bandit problems.

In this talk, a Booking.com Machine Learning Scientist will introduce how to recognize Bandit problems and which algorithms exist to solve them. At the end of the talk, you should have a basic understanding of Bandit Algorithms and Counterfactual Learning, techniques that are applicable in a wide range of problem statements.

This lecture is worth 1 colloquium point. You don't need to sign up for the lecture.