Description du projet
S’attaquer à l’agrégation des préférences dans les systèmes multi-agents
Dans le domaine complexe des systèmes multi-agents (SMA), le défi de l’agrégation de préférences individuelles conflictuelles se pose avec acuité, en particulier face à des agents égoïstes. La recherche de résultats socialement souhaitables devient encore plus difficile lorsque les agents peuvent manipuler le système en mentant sur leurs préférences. Cet obstacle s’étend à diverses applications du monde réel, des protocoles de répartition équitable aux systèmes de vote sécurisés. Avec le soutien du programme Actions Marie Skłodowska-Curie, le projet SEAL s’appuie sur la logique stratégique pour forger un cadre logique permettant de concevoir et de vérifier formellement des mécanismes qui naviguent dans les complexités du choix social au sein des SMA. Le projet vise à renforcer la confiance dans les mécanismes générés par les machines. SEAL promet de révolutionner le paysage du choix social computationnel.
Objectif
"The design and evaluation of mechanisms for aggregating preferences is a central problem in Multi-Agent Systems (MAS). In such setting, we need to be able to aggregate individual preferences, which are conflicting when agents are self-interested. More importantly, the mechanism should choose a socially desirable (or ""good"") outcome and reach an equilibrium despite the fact that agents can lie about their preferences. The real-world applications of designing and verifying mechanisms for social choice are manifold, including fair division protocols, secure voting, and truth-tracking via approval voting. Although logic-based languages have been widely used for verification and synthesis of MAS, the use of formal methods for reasoning about auctions under strategic behavior as well as automated mechanism design has not been much explored yet. An advantage in adopting such perspective lies in the high expressivity and generality of logics for strategic reasoning. Moreover, by relying on precise semantics, formal methods provide tools for rigorously analyzing the correctness of systems, which is important to improve trust in mechanisms generated by machines. This project aims to design a logical framework based on Strategy Logic (SL) for formally verifying and designing mechanisms for social choice. More specifically, we aim at (i) proposing an approach addressing the probabilistic setting (with Bayesian information, stochastic transitions and mixed strategies); (ii) identifying fragments of SL that enjoy both good complexity and satisfying expressive power for being applied to classes of mechanisms; (iii) modeling and reasoning about relevant problems from the state-of-the-art in computational social choice using the proposed logical framework; and (iv) methodically studying the obtained fragments in relation to the expressivity, model-checking and satisfiability problems."
Champ scientifique
Mots‑clés
Programme(s)
- HORIZON.1.2 - Marie Skłodowska-Curie Actions (MSCA) Main Programme
Régime de financement
HORIZON-TMA-MSCA-PF-EF - HORIZON TMA MSCA Postdoctoral Fellowships - European FellowshipsCoordinateur
80138 Napoli
Italie