Summary
"Market designers study how to set the ""rules of a marketplace"" such that the market works well. However, markets are getting increasingly complex such that designing good market mechanisms ""by hand"" is often infeasible, in particular when certain design desiderata (such as efficiency, strategyproofness, or fairness) are in conflict with each other. Moreover, human agents are boundedly-rational: already in small domains, they are best modeled as having incomplete preferences, because they may only know a ranking or the values of their top choices. In combinatorial domains, the number of choices grows exponentially, such that it quickly becomes impossible for an agent to report its full valuation, even if it had complete preferences. In this ERC grant proposal, we propose to combine techniques from ""machine learning"" with ""market design"" to address these challenges.
First, we propose to develop a new, automated approach to design mechanisms with the help of machine learning (ML). In contrast to prior ML-based automated mechanism design work, we explicitly aim to train the ML algorithm to exploit regularities in the mechanism design space. Second, we propose to study the ""design of machine learning-based mechanisms."" These are mechanisms that use machine learning internally to achieve good efficiency and incentives even when agents have incomplete knowledge about their own preferences.
In addition to pushing the scientific boundaries of market design research, this ERC project will also have an immediate impact on practical market design. We will apply our techniques in two different settings: (1) for the design of combinatorial spectrum auctions, a multi-billion dollar domain; and (2) for the design of school choice matching markets, which are used to match millions of students to high school every year.
"
First, we propose to develop a new, automated approach to design mechanisms with the help of machine learning (ML). In contrast to prior ML-based automated mechanism design work, we explicitly aim to train the ML algorithm to exploit regularities in the mechanism design space. Second, we propose to study the ""design of machine learning-based mechanisms."" These are mechanisms that use machine learning internally to achieve good efficiency and incentives even when agents have incomplete knowledge about their own preferences.
In addition to pushing the scientific boundaries of market design research, this ERC project will also have an immediate impact on practical market design. We will apply our techniques in two different settings: (1) for the design of combinatorial spectrum auctions, a multi-billion dollar domain; and (2) for the design of school choice matching markets, which are used to match millions of students to high school every year.
"
Unfold all
/
Fold all
More information & hyperlinks
| Web resources: | https://cordis.europa.eu/project/id/805542 |
| Start date: | 01-12-2018 |
| End date: | 30-11-2023 |
| Total budget - Public funding: | 1 375 000,00 Euro - 1 375 000,00 Euro |
Cordis data
Original description
"Market designers study how to set the ""rules of a marketplace"" such that the market works well. However, markets are getting increasingly complex such that designing good market mechanisms ""by hand"" is often infeasible, in particular when certain design desiderata (such as efficiency, strategyproofness, or fairness) are in conflict with each other. Moreover, human agents are boundedly-rational: already in small domains, they are best modeled as having incomplete preferences, because they may only know a ranking or the values of their top choices. In combinatorial domains, the number of choices grows exponentially, such that it quickly becomes impossible for an agent to report its full valuation, even if it had complete preferences. In this ERC grant proposal, we propose to combine techniques from ""machine learning"" with ""market design"" to address these challenges.First, we propose to develop a new, automated approach to design mechanisms with the help of machine learning (ML). In contrast to prior ML-based automated mechanism design work, we explicitly aim to train the ML algorithm to exploit regularities in the mechanism design space. Second, we propose to study the ""design of machine learning-based mechanisms."" These are mechanisms that use machine learning internally to achieve good efficiency and incentives even when agents have incomplete knowledge about their own preferences.
In addition to pushing the scientific boundaries of market design research, this ERC project will also have an immediate impact on practical market design. We will apply our techniques in two different settings: (1) for the design of combinatorial spectrum auctions, a multi-billion dollar domain; and (2) for the design of school choice matching markets, which are used to match millions of students to high school every year.
"
Status
CLOSEDCall topic
ERC-2018-STGUpdate Date
27-04-2024
Images
No images available.
Geographical location(s)
