Three-Way Group Decision Via Joint Classification and Consensus

In three-way group decision-making based on the minimum risk, the collective consensus on the evaluation of basic loss functions is a central issue [13]. However, focusing solely on evaluation information does not ensure the classification quality of three-way decisions. To balance the consensus and decision quality, this chapter elaborates a joint learning process for three-way group decision-making by constructing a two-stage group consensus method of Ref. [15]. Considering supervised learning, Stage 1 establishes a minimum decision error optimization model (MDEOM) to learn the optimal parameters for three-way decisions and calculate reference values for decision loss. We then solve the MDEOM using a particle swarm optimization (PSO) algorithm. In Stage 2, adjusted decision losses are calculated using the minimum decision loss difference consensus model (MDLDCM), which guides the consensus adjustment of loss functions and improves the decision quality of collective wisdom. Finally, a series of experiments are discussed to illustrate the implications of three-way group decision-making with the two-stage group consensus method based on the results of Ref. [15].