Taming Dilation in Imprecise Pooling

Jonas KargeJonas Karge

Download

Jonas Karge
Taming Dilation in Imprecise Pooling
Proceedings of the International Conference on Principles and Practice of Multi-Agent Systems (PRIMA), to appear

Details
Bibtex

KurzfassungAbstract
If an agent’s belief in a proposition is represented by imprecise probabilities, i.e. intervals of probability values, a phenomenon called dilation can occur, where updating the agent’s belief with a new observation can only widen the probability interval, thus making the agent more uncertain, regardless of the observation made. Similar to standard updating, dilation can also occur in the context of imprecise opinion pooling, where the imprecise beliefs of multiple agents are aggregated. In this work, we provide the first formal investigation of dilation and its counterpart, contraction, in the context of imprecise opinion pooling. To this end, we use a recently defined voting rule, Voting for Bins (VfB), as a means to handle dilation and contraction, consistent with intuitions about the quality of additional opinions. VfB, inspired by the Condorcet Jury Theorem (CJT), is extended to account for correlation by an opinion leader. This model is further generalized to account for average correlation.
Projekt:Project: SECAI
Forschungsgruppe:Research Group: Computational LogicComputational Logic

@inproceedings{K2024,
  author    = {Jonas Karge},
  title     = {Taming Dilation in Imprecise Pooling},
  booktitle = {Proceedings of the International Conference on Principles and
               Practice of Multi-Agent Systems (PRIMA)},
  year      = {2024}
}