EURO 2024 Copenhagen
Abstract Submission

EURO-Online login

2031. Approximate computation of the Shapley value in cooperative games on transportation networks via machine-learning techniques

Invited abstract in session WD-43: Recent advances on Variational Inequalities and Equilibrium Problems III , stream Variational Inequalities and Equilibrium Problems: From Theoretical Advances to Real World Applications.

Wednesday, 14:30-16:00
Room: 99 (building: 306)

Authors (first author is the speaker)

1. Mauro Passacantando
Department of Business and Law, University of Milano-Bicocca
2. Giorgio Gnecco
IMT - School for Advanced Studies, Lucca
3. Yuval Hadas
Department of Management, Bar-Ilan University
4. Marcello Sanguineti
DIBRIS, University of Genoa

Abstract

To quantify the relative importance of nodes in a network, one possible approach uses the concept of centrality. However, classical measures of centrality do not consider how the network behaves if any subset of nodes is removed from the network and, when applied to transportation networks, issues such as arc capacity, route choice, and congestion are not taken into account. Alternative measures of centrality based on game theory are more effective, such as those based on the Shapley value. The Shapley value is a well-established concept in cooperative game theory, used as a metric for assessing the significance of each player in a transferable utility game. Recently, it has found application in gauging the importance of individual arcs or nodes within transportation networks. However, the exact computation of the Shapley value is often computationally expensive, particularly in the context of extensive networks. Here, we approximate the Shapley value in a transferable utility game defined on a network, wherein the network’s characteristics are parameterized by a variable of interest (e.g., the traffic demand). Smoothness properties of the Shapley value as a function of the parameter are investigated and exploited to motivate using machine-learning techniques for its approximate computation.

Keywords

Status: accepted

Back to the list of papers