On extension of 2-copulas for information fusion

Copulas are very applicable tools from a statistical point of view. Their capability to model joint distribution makes them a useful tool for statisticians. Yet, they have been applied rarely in multi-criteria decision aiding (MCDA). This paper introduces a novel family of aggregation functions based on 2-copulas for MCDA. More specifically, the proposed copula-based method is applied to construct the popular Choquet integral (CI) and attitudinal Choquet integral (ACI). The forms obtained are utilized to develop a machine learning model to explain choices in the human decision-making situations. Exemplary pairwise preferences are provided as training information for the proposed model. Our approach effectively represents the dependencies between criteria under uncertainty. Experimental results on multiple real-world datasets demonstrate that our approach significantly outperforms popular MCDA methods, such as RBF-kernel, Tchebycheff, etc.