Title: Design of Combinatorial Auctions Using Machine Learning-Based Bidding
Abstract:
Combinatorial auctions are at the core of Economics and Computation research due to the large computational resources that they require.
Bidding languages are often used to overcome this issue. However, each of them imposes some structure in bidders’ valuation. This structure may be hard to determine and may constrain the expressiveness of the bidders, leading then to inefficiencies.
In this talk new bidding languages based on machine learning techniques will be presented. These bidding languages can exploit some structures inferred from past auctions data or provided by the auction designer without constraining the agents’ expressiveness
