Researches On Attribute Reduction Of Utility Three-way Decisions Model With Rough Sets

The three-way decisions model is a cost-sensitive decision learning method that introduces Bayesian risk decision concept and conditional probability formula in rough set theory by Professor Yao Yiyu to analyze and summarize the decision behavior.The threeway decisions model have three decision modes: accept decision,reject decision and delay decision.Summarizing decision rules through this classification decision model is suitable for solving actual decision problems,and provides help for decision makers to summarize decision rules and make optimal decision choices.The three-way decisions model also have many applications in the field of artificial intelligence: such as license plate number image recognition,spam classification and filtering,and clinical expert decision systems.The application of the three-way decisions model is gradually gaining attention from researchers.The classic three-way decisions model is a cost-sensitive model that summarizes the decision rules based on objective expectation risk.Some scholars combined the wishes of decision-makers and used utility to express the subjective attitude of them,and proposed a utility-based three-way decisions model.However,due to the existence of multiple types of data in the actual decision information systems,the existing utility-based three-way decisions model cannot simultaneously handle real-valued,Boolean,and categorical data,and cannot solve the n-class of the classification problem.the expected utility theory used in this model is based on the absolute rational decision-making of human beings,and only uses a single utility value to describe the psychological satisfaction of the decision-maker,which contradicts the behavior of the decision-maker's emotional decision-making in practice.This paper combines a utility concept in a hybrid decision system to propose a new fuzzy utility three-way decisions model,and in this model proposes a heuristic algorithm attribute reduction for maximum utility and positive region distribution preservation.The main construction process of fuzzy utility three-way decisions model is: Constructing fuzzy conditional probability in hybrid information system,combining utility concept to generate thresholds for different decision classes,and finally verifying the algorithm proposed in this paper with respect to utility three-way decisions model improves the accuracy of model classification.The value function in the prospect theory increases the decision reference point,balances the loss and the benefit to define the satisfaction degree of the decision-makers,and also considers the decision loss and the decision benefit,relative to the single utility value description in the expected utility theory.For the behavior of decision-makers,prospect theory is more suitable for analysis of decision-making behavior with three-way decisions models.In order to solve the above problems,this paper introduces the prospect theory into the multi-class three-way decisions model based on the concept of the multiclass three-way decisions model,and uses the prospect value function to represent people's benefits and costs in actual life,construct a classification value matrix,and propose a multi-class three-way decisions model based on prospect theory.In this model,based on the prospect theory and the classification value matrix,this paper proposes the goal of maximum prospect value reduction.Under this goal,a heuristic algorithm of forward greedy maximum prospect value attribute reduction and backward greedy maximum value heuristic attribute reduction algorithm are proposed.