BundleFlow: Deep Menus for Combinatorial Auctions by Diffusion-Based Optimization

TL;DR

BundleFlow introduces a novel diffusion-based deep learning approach to optimize combinatorial auction menus, significantly improving revenue and training efficiency in high-dimensional item spaces.

Contribution

The paper presents BundleFlow, a diffusion-inspired method for generating bundle distributions in combinatorial auctions, addressing exponential complexity and outperforming existing baselines.

Findings

Achieves 1.11-2.23x higher revenue than baselines.

Scales to 150 items in combinatorial auctions.

Reduces training iterations by up to 9.5x and training time by 80%.

Abstract

Differentiable economics -- the use of deep learning for auction design -- has driven progress in the automated design of multi-item auctions with additive or unit-demand valuations. However, little progress has been made for optimal combinatorial auctions (CAs), even for the single bidder case, because we need to overcome the challenge of the bundle space growing exponentially with the number of items. For example, when learning a menu of allocation-price choices for a bidder in a CA, each menu element needs to efficiently and flexibly specify a probability distribution on bundles. In this paper, we solve this problem in the single-bidder CA setting by generating a bundle distribution through an ordinary differential equation (ODE) applied to a tractable initial distribution, drawing inspiration from generative models, especially score-based diffusion models and continuous normalizing flow. Our method, BundleFlow, uses deep learning to find suitable ODE-based transforms of initial distributions, one transform for each menu element, so that the overall menu achieves high expected revenue. Our method achieves 1.11$-$2.23$\times$ higher revenue compared with automated mechanism design baselines on the single-bidder version of CATS, a standard CA testbed, and scales to problems with up to 150 items. Relative to a baseline that also learns allocations in menu elements, our method reduces the training iterations by 3.6$-$9.5$\times$ and cuts training time by about 80% in settings with 50 and 100 items.