Combinatorial Auctions via Machine Learning-based Preference Elicitation

Brero, Gianluca; Lubin, Benjamin; Seuken, Sven

doi:10.24963/ijcai.2018/18

Cited by 19 publications

(36 citation statements)

References 9 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…As part of their algorithm, they used kernelized support vector regressions (SVRs) to learn the nonlinear value functions of bidders. Recently, Brero, Lubin, and Seuken (2019) showed that their MLbased ICA achieves even higher efficiency than the CCA. However, because of runtime complexity issues, Brero, Lubin, and Seuken (2018; focused on SVRs with linear and quadratic kernels.…”

Section: Machine Learning and Mechanism Designmentioning

confidence: 99%

“…Recently, Brero, Lubin, and Seuken (2019) showed that their MLbased ICA achieves even higher efficiency than the CCA. However, because of runtime complexity issues, Brero, Lubin, and Seuken (2018; focused on SVRs with linear and quadratic kernels. This leaves room for improvement, since bidders' valuations can have more complex structures than can be captured by linear or quadratic kernels.…”

Section: Machine Learning and Mechanism Designmentioning

confidence: 99%

“…represents bidder i's true value for bundle x. 1 We compare our DNN-powered ICA against the mechanism described in (Brero, Lubin, and Seuken 2018) because, when we wrote this paper, (Brero, Lubin, and Seuken 2019) was not available yet. We slightly adopt the notation and use Bi instead of ϑi.…”

Section: Iterative Combinatorial Auctionmentioning

confidence: 99%

See 2 more Smart Citations

Deep Learning—Powered Iterative Combinatorial Auctions

Weissteiner

Seuken

2020

AAAI

Self Cite

View full text Add to dashboard Cite

In this paper, we study the design of deep learning-powered iterative combinatorial auctions (ICAs). We build on prior work where preference elicitation was done via kernelized support vector regressions (SVRs). However, the SVR-based approach has limitations because it requires solving a machine learning (ML)-based winner determination problem (WDP). With expressive kernels (like gaussians), the ML-based WDP cannot be solved for large domains. While linear or quadratic kernels have better computational scalability, these kernels have limited expressiveness. In this work, we address these shortcomings by using deep neural networks (DNNs) instead of SVRs. We first show how the DNN-based WDP can be reformulated into a mixed integer program (MIP). Second, we experimentally compare the prediction performance of DNNs against SVRs. Third, we present experimental evaluations in two medium-sized domains which show that even ICAs based on relatively small-sized DNNs lead to higher economic efficiency than ICAs based on kernelized SVRs. Finally, we show that our DNN-powered ICA also scales well to very large CA domains.

show abstract

Section: Machine Learning and Mechanism Designmentioning

confidence: 99%

Section: Machine Learning and Mechanism Designmentioning

confidence: 99%

See 1 more Smart Citation

Deep Learning—Powered Iterative Combinatorial Auctions

Weissteiner

Seuken

2020

AAAI

Self Cite

View full text Add to dashboard Cite

show abstract

“…One major difficulty arises from the fact that the true valuation functions v i are unknown to the auctioneer and can only be accessed through a limited amount of queries (typically less than 500 queries per bidder) called preference elicitation [48]. Machine learning based preference elicitation approaches overcome this issue by approximating the valuation functions by parametric functions, e.g., polynomials of degree two [48] or Gaussian processes [49]. The estimated parameters of these approximations are adaptively refined using a suitable querying strategy.…”

Section: Samplingmentioning

confidence: 99%

A Discrete Signal Processing Framework for Set Functions

Püschel

2018

2018 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP)

View full text Add to dashboard Cite

Set functions are functions (or signals) indexed by the power set (set of all subsets) of a finite set N . They are ubiquitous in many application domains. For example, they are equivalent to node-or edge-weighted hypergraphs and to cooperative games in game theory. Further, the subclass of submodular functions occurs in many optimization and machine learning problems. In this paper, we derive discrete-set signal processing (SP), a shiftinvariant linear signal processing framework for set functions. Discrete-set SP provides suitable definitions of shift, shift-invariant systems, convolution, Fourier transform, frequency response, and other SP concepts. Different variants are possible due to different possible shifts. Discrete-set SP is inherently different from graph SP as it distinguishes the neighbors of an index A ⊆ N , i.e., those with one elements more or less by providing n = |N | shifts. Finally, we show three prototypical applications and experiments with discrete-set SP including compression in submodular function optimization, sampling for preference elicitation in auctions, and novel power set neural networks.

show abstract

“…and have been a central topic in the study of Multi-Agent Systems. They have also experienced a recent interest in the AI community with works employing ML algorithms to overcome standard complexity problems (e.g., [5,6]).…”

Section: Introductionmentioning

confidence: 99%

A Bridge between Liquid and Social Welfare in Combinatorial Auctions with Submodular Bidders

Fotakis

Lotidis

Podimata

2019

AAAI

View full text Add to dashboard Cite

We study incentive compatible mechanisms for Combinatorial Auctions where the bidders have submodular (or XOS) valuations and are budget-constrained. Our objective is to maximize the liquid welfare, a notion of efficiency for budget-constrained bidders introduced by Dobzinski and Paes Leme (2014). We show that some of the known truthful mechanisms that bestapproximate the social welfare for Combinatorial Auctions with submodular bidders through demand query oracles can be adapted, so that they retain truthfulness and achieve asymptotically the same approximation guarantees for the liquid welfare. More specifically, for the problem of optimizing the liquid welfare in Combinatorial Auctions with submodular bidders, we obtain a universally truthful randomized O(log m)-approximate mechanism, where m is the number of items, by adapting the mechanism of Krysta and Vöcking (2012).Additionally, motivated by large market assumptions often used in mechanism design, we introduce a notion of competitive markets and show that in such markets, liquid welfare can be approximated within a constant factor by a randomized universally truthful mechanism. Finally, in the Bayesian setting, we obtain a truthful O(1)-approximate mechanism for the case where bidder valuations are generated as independent samples from a known distribution, by adapting the results of Feldman, Gravin and Lucier (2014).

show abstract

Combinatorial Auctions via Machine Learning-based Preference Elicitation

Cited by 19 publications

References 9 publications

Deep Learning—Powered Iterative Combinatorial Auctions

Deep Learning—Powered Iterative Combinatorial Auctions

A Discrete Signal Processing Framework for Set Functions

A Bridge between Liquid and Social Welfare in Combinatorial Auctions with Submodular Bidders

Contact Info

Product

Resources

About