14:00 – 15:00 Sofia Villar

Title: Connecting Multi-armed Bandit Problems, Index Policies and Clinical Trial Design

Abstract: In this talk I will present some high level ideas from my own research establishing a connection between the multi-armed bandit problem (as it was studied in the Operations Research literature), its computationally efficient solution by means of priority index heuristics and its application to designing patient allocation rules for adaptive clinical trials.

15:30 – 16:30 Nicolo Cesa-Bianchi

Title: Trading-Off Payments and Accuracy in Online Classification

Abstract: We investigate methods for the online aggregation of experts in classification tasks where, before making their prediction, each expert must be paid. The amount that we pay each expert directly influences the accuracy of their prediction through some unknown productivity function. In each round, the learner must decide how much to pay each expert and then make a prediction. They incur a cost equal to a weighted sum of the prediction error and upfront payments for all experts. We introduce an online learning algorithm and analyse its total cost compared to that of a predictor which knows the productivity of all experts in advance. In order to achieve this result, we combine Lipschitz bandits and online classification with surrogate losses.

Joint work with: Dirk van der Hoeven, Ciara Pike-Burke, Hao Qiu

Refreshments available between 15:00 – 15:30