Fuzzy Inference Based A Posterior Decision-Making for Multi-Objective Diet Optimization Problem

Cumali TÜRKMENOĞLU,Ayşe Şima UYAR,Berna KİRAZ
DOI: https://doi.org/10.31590/ejosat.1216648
2022-12-21
European Journal of Science and Technology
Abstract:We propose a Mamdani-Type Fuzzy Inference based posterior decision-making approach to multi-objective diet optimization problem. We optimize the multi-objective diet problem with evolutionary algorithms that result in tens/hundreds of non-dominated solutions which is too large to pick one of them by the decision-maker. Even though all the solutions are optimized for all the objectives simultaneously, not all objective functions may be equally important to a user and, also their importance may change for that user over time. Our main goal is to develop an applicable method for representing and incorporating a decision maker's (DM) instant preferences for objectives into decision-making stage. The FIS based decision making can guide users to decide on the most suitable menus. User's instant preferences for each objective form rule sets. Using Mamdani type FIS in the post-decision process of the multi-objective diet problem is a novel contribution. A desirability measure is calculated by using rule sets and membership functions considering the objective values, and based on the desirability measure the most preferred menu(s) are provided to the user. Our method can direct the DM to the region of interest in the search space of the multi-objective diet problem. Thus, the daily menu suggestions become more applicable, practical, and desirable for the users.
What problem does this paper attempt to address?