Learning Market Equilibria Preserving Statistical Privacy Using Performative Prediction

Le Cadre, Hélène; Datar, Mandar; Guckert, Mathis; Altman, Eitan

Document type :

Pré-publication ou Document de travail

Title :

Learning Market Equilibria Preserving Statistical Privacy Using Performative Prediction

Author(s) :

Le Cadre, Hélène [Auteur]
Integrated Optimization with Complex Structure [INOCS]
Datar, Mandar [Auteur]
Laboratoire Informatique d'Avignon [LIA]
Guckert, Mathis [Auteur]
Integrated Optimization with Complex Structure [INOCS]
Altman, Eitan [Auteur]
Network Engineering and Operations [NEO]
Laboratoire Informatique d'Avignon [LIA]

English keyword(s) :

Learning Game
Mechanism Design
Performatively Stable Equilibrium
Statistical Privacy

HAL domain(s) :

Mathématiques [math]/Optimisation et contrôle [math.OC]

English abstract : [en]

We consider a peer-to-peer electricity market modeled as a network game, where End Users (EUs) minimize their cost by computing their demand and generation while satisfying a set of local and coupling constraints. Their ...
Show more >We consider a peer-to-peer electricity market modeled as a network game, where End Users (EUs) minimize their cost by computing their demand and generation while satisfying a set of local and coupling constraints. Their nominal demand constitutes sensitive information, that they might want to keep private. We prove that the network game admits a unique Variational Equilibrium, which depends on the private information of all the EUs. A data aggregator (DA) is introduced, which aims to learn the EUs' private information, while remunerating them depending on the quality of the readings they report to the DA. The EUs might have incentives to report biased and noisy readings to preserve their privacy. Relying on performative prediction, we define a decision-dependent game G stoch , to explicitly take into account the shift caused by the EUs' strategic information on their strategies and market equilibria. To compute market equilibria solutions of G stoch , two variants of the Repeated Stochastic Gradient Method (RSGM) and a two-timescale stochastic approximation algorithm are proposed. We prove the convergence of each algorithm. Finally, the algorithms performance is assessed on a numerical example, by comparing the achieved efficiency loss, privacy preservation capabilities, convergence rates, and EUs' utility functions at equilibrium. The results highlight the benefits for the EUs to model performative effects.Show less >

Language :

Anglais

Collections :

Centre de Recherche en Informatique, Signal et Automatique de Lille (CRIStAL) - UMR 9189

Source :

Harvested from HAL

Files

document
Open access
Access the document

Privacy_Contract%20%2814%29.pdf
Open access
Access the document

Learning Market Equilibria Preserving ... BibTeX CSV Excel RIS

Files

Learning Market Equilibria Preserving ...

BibTeX

CSV

Excel

RIS