Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Efficient Preference Elicitation in Iterative Combinatorial Auctions with Many Participants

Ryota Maruo, Hisashi Kashima

TL;DR

A multi-task learning method to learn valuation functions more efficiently that achieves higher efficiency than existing methods, especially in scenarios with many bidders and items but a limited number of queries.

Abstract

We study the problem of achieving high efficiency in iterative combinatorial auctions (ICAs). ICAs are a kind of combinatorial auction where the auctioneer interacts with bidders to gather their valuation information using a limited number of queries, aiming for efficient allocation. Preference elicitation, a process that incrementally asks bidders to value bundles while refining the outcome allocation, is a commonly used technique in ICAs. Recently, the integration of machine learning (ML) into ICAs has significantly improved preference elicitation. This approach employs ML models that match the number of bidders, estimating each bidder's valuation functions based on their reported valuations. However, most current studies train a separate model for each bidder, which can be inefficient when there are numerous bidders with similar valuation functions and a limited number of available queries. In this study, we introduce a multi-task learning method to learn valuation functions more efficiently. Specifically, we propose to share model parameters during training to grasp the intrinsic relationships between valuations. We assess the performance of our method using a spectrum auction simulator. The findings demonstrate that our method achieves higher efficiency than existing methods, especially in scenarios with many bidders and items but a limited number of queries.

Efficient Preference Elicitation in Iterative Combinatorial Auctions with Many Participants

TL;DR

A multi-task learning method to learn valuation functions more efficiently that achieves higher efficiency than existing methods, especially in scenarios with many bidders and items but a limited number of queries.

Abstract

We study the problem of achieving high efficiency in iterative combinatorial auctions (ICAs). ICAs are a kind of combinatorial auction where the auctioneer interacts with bidders to gather their valuation information using a limited number of queries, aiming for efficient allocation. Preference elicitation, a process that incrementally asks bidders to value bundles while refining the outcome allocation, is a commonly used technique in ICAs. Recently, the integration of machine learning (ML) into ICAs has significantly improved preference elicitation. This approach employs ML models that match the number of bidders, estimating each bidder's valuation functions based on their reported valuations. However, most current studies train a separate model for each bidder, which can be inefficient when there are numerous bidders with similar valuation functions and a limited number of available queries. In this study, we introduce a multi-task learning method to learn valuation functions more efficiently. Specifically, we propose to share model parameters during training to grasp the intrinsic relationships between valuations. We assess the performance of our method using a spectrum auction simulator. The findings demonstrate that our method achieves higher efficiency than existing methods, especially in scenarios with many bidders and items but a limited number of queries.
Paper Structure (23 sections, 7 equations, 2 figures, 1 table, 2 algorithms)

This paper contains 23 sections, 7 equations, 2 figures, 1 table, 2 algorithms.

Table of Contents

  1. Introduction
  2. Related Work
  3. Preference Elicitation in Combinatorial Auctions
  4. Automated Mechanism Designs in Auctions
  5. Preliminaries
  6. Problem Setting
  7. ML-powered ICA
  8. Proposed Method
  9. Soft Parameter-Sharing
  10. ID Injection
  11. Experimental Result
  12. Spectrum Auction Test Suite (SATS)
  13. Experimental Setup
  14. MLCA Configuration
  15. Model Architectures
  16. ...and 8 more sections

Figures (2)

  • Figure 1: Mean efficiency in the $(3l, 4l, 3l) (l=1,\dots, 5)$ settings. The horizontal axis represents the number of (local, regional, national) bidders, respectively, while the vertical axis shows the efficiency. The left panel displays the efficiency results for 98 items and the right panel for 196 items. Both figures share the same legend.
  • Figure 2: MAPE results for all settings. The horizontal axis represents the number of rounds $k$ in MLCA, and the vertical axis denotes the MAPE. The first row of five figures presents the results for 98 items with $(3l,4l,3l) (l=1,\dots, 5)$ bidders, while the second row of five figures displays the results for 196 items. All figures share the legend presented in the first one.