Aggregating computational algorithms and human decision-making preferences in multi-agent settings
Abstract
We propose to develop models for multi-agent voting aggregation in complex socio-technical systems. While the study of human bias is already a well-established area research, understanding human bias in complex socio-technical system remains a challenging task. It is because algorithms produce bias that is most often not understandable to humans and hidden from them, and because studies of voting in systems where algorithms and humans participate jointly are still not well studied. The objective of this proposal is to provide scientific foundations for developing algorithms and methods for multi-agent voting aggregation. To prove the quality of the models we plan to apply the theoretical insights for solving problems for identification of experts and optimal team composition in multi-agent settings, as well as in participatory budgeting. The project objectives will be achieved by addressing four aspects (1) Modeling utilities of human decision-makers in multi-agent settings, (2) Decomposing the bias structure in decision-making algorithms, (3) Application: Identification of experts and optimal composition of teams based on human and algorithm voting, (4) Application: Participatory budgeting based on human and algorithm voting. The potential relevance of this research for US Navy is tremendous. As decision-making is of utmost importance for the military, optimizing the way how people collaborate and make decisions can improve the operations of the US Navy (structures can range from structured organizations such as command systems to the decentralized and elusive adversary organizations such as insurgent and terrorist groups). The planned project outcomes are several major journal and conference papers.
Document Details
- Document Type
- DoD Grant Award
- Publication Date
- Apr 25, 2019
- Source ID
- N629091912008
Entities
People
- Boris Delibašić
Organizations
- Office of Naval Research
- United States Navy