The Department would like to congratulate to Professor Shipra Agrawal for earning two awards for her research. Professor Agrawal was awarded the Amazon Research Award (ARA) for her proposal on “New Algorithmic Approaches for Reinforcement Learning, with Application to Integer Programming” and a Google Research Awardfor her proposal on "Robust Auction Mechanisms under Heterogeneous Buyer Behavior".
Amazon Research Award: New Algorithmic Approaches for Reinforcement Learning, with Application to Integer Programming
For this proposal, Professor Agrawal is investigating how new reinforcement learning algorithms can achieve scalable solutions for large and complex reinforcement learning problems where there is a lack of a system dynamic simulator (e.g. unexplored terrains or unpredictable environments) and there exists a complex action space (e.g. gaming applications that involve controls and navigation). Her focus is a specific problem-solving application for building a reinforcement learning agent that can learn to solve Integer Programs (IP). [More]
Google Research Award: Robust Auction Mechanisms under Heterogeneous Buyer Behavior
In this project, Professor Agrawal is studying new robust dynamic mechanisms for optimizing revenue in repeated auctions. In these scenarios, the seller is interested in the buyer's variety of learning and planning behaviors in response to his or her actions. Research in this area is important as it provides better insights and business actions to Google's display ad exchange, which includes a large fraction of advertisements that are sold through repeated auctions. [More]