Distributed decision making
In today’s complex world, decision-making often involves multiple people or artificial agents, each with their own goals and limits. Distributed solutions make complex dynamic decision making tasks feasible. The research focuses on how these imperfect decision-makers can effectively collaborate, negotiate or compete to achieve better outcomes.
We have built on the theory designed for individual decision-makers and expanded it to accommodate a group of agents working together. Our key areas of focus include: automatic elicitation and harmonization of preferences; knowledge fusion; cooperation and negotiation.
Our results are built around three key components:
DM preference elicitation
Preference elicitation for dynamic decision-making involves gathering and understanding the preferences of decision-makers in real-time, adapting to changing circumstances. This process is essential for developing effective strategies that align with individuals’ goals, allowing for informed choices in complex, evolving environments.
An automated and optimized solution within the FPD framework has been developed to accommodate agents’ fragmented knowledge and unquantified multi-attribute desires. The solution unifies the handling of beliefs and preferences, enabling the combination of multiple attributes while minimizing conflicts.
Automatic mapping of decision goals
We offer robust and well-grounded methods for translating available informal or incomplete knowledge and decision goals into probabilities, while accounting for the limited resources of real agents, such as expertise, time for deliberation, data availability, computing power, and energy requirements .
Cooperation and indirect dynamic negotiation
To foster cooperation in a distributed setting, we have developed a fusion algorithm that optimally leverages privately designed forecasting probability densities from neighboring agents to enhance the focal agent’s information.
In addition, we created a trust-based knowledge-sharing algorithm that allows the focal agent to estimate the trustworthiness of information received from its neighbors, effectively addressing both gradually and abruptly fluctuating credibility. This dual approach not only enriches the decision-making process but also strengthens collaborative efforts among agents.
Related publications:
- Kárný, M. Fully Probabilistic Design Unifies and Supports Dynamic Decision Making Under Uncertainty. Information Sciences. 2020.
- Guy T. V., Homolová Jitka, Gaj A. : Indirect Dynamic Negotiation in the Nash Demand Game , IEEE Access 10(1):105008-21, 2022
- Kárný, M., Siváková, T. Model-based preference quantification. Automatica. 2023, 156(1).
- Kárný, M., Guy, T.V., Preference Elicitation within Framework of Fully Probabilistic Design of Decision Strategies, IFAC-Papers online 2019, 52(29):239-244.
- Kárný, M., Hůla, F. Fusion of Probabilistic Unreliable Indirect Information into Estimation Serving to Decision Making. Int.J.Machine Learning & Cybernetics. 2021, 12(12).
- Kárný M., Karlík D.: Trust Estimation in Forecasting-Based Knowledge Fusion, Proc. of BNAIC – Benelux Conference on Artificial Intelligence and 30th Belgian-Dutch Conference on Machine Learning 2021.
- Kuklišová Pavelková L., Jirsa L., Quinn A.: Fully probabilistic design for knowledge fusion between Bayesian filters under uniform disturbances, Knowledge-Based System 238, 2022
- Homolová J., Černecka A., Guy T.V., Kárný M.: Affective Decision-Making in Ultimatum Game: Responder Case, Multi-Agent Systems: 16th European Conference, EUMAS 2018, Revised Selected Papers, Eds: Slavkovik Marija, Eumas, Bergen, NO, 20181206, 2019
- Ruman M., Hůla F., Kárný M., Guy T.V.: Deliberation-aware Responder in Multi-Proposer Ultimatum Game, Artificial Neural Networks and Machine Learning – ICANN 2016, 230-237, 2016
Kárný M., Guy T. V., Bodini A., Ruggeri F.: Cooperation via sharing of probabilistic information, International Journal of Computational Intelligence Studies, 139-162, 2009