Challenges in multi-aspect decision making: integrating machine learning and operations research techniques

• Grant number: 23/04159-6
• Support Opportunities: Scholarships in Brazil - Post-Doctoral
• Start date: June 01, 2023
• End date: January 31, 2024
• Field of knowledge: Engineering - Production Engineering - Operational Research
• Principal Investigator: Leonardo Tomazeli Duarte
• Grantee: Betania Silva Carneiro Campello
• Host Institution: Faculdade de Ciências Aplicadas (FCA). Universidade Estadual de Campinas (UNICAMP). Limeira , SP, Brazil
• Associated research grant(s): 24/03035-4 - 27th International conference on Multiple Criteria Decision Making (MCDM 2024), AR.EXT

Abstract

Decisions in various domains, including personal, social, and governmental, are typically made by considering multiple relevant aspects. These aspects may include decision criteria, conflicting objectives, or even multiple modes of information, such as multimedia. Decision-making involving multiple aspects often presents challenges as it requires significant effort in data analysis and relies on the use of appropriate mathematical techniques and models for decision support. In the field of decision science, areas such as operations research (OR) and machine learning (ML) are efficient in dealing with this large amount of information, helping to make decisions simpler and more complete. Both OR and ML methods present some advantages and disadvantages in decision support. A more current trend is the development of hybrid methods of OR and ML, aiming to obtain procedures that are more focused on ethics, easy to understand, and complete. This research project proposes new hybrid techniques of OR and ML to support complex decision-making processes that involve the analysis of multiple aspects. We approach the problem with two fronts. First, we intend to deal with decisions involving multiple criteria and dimensions. The second front deals with multimodal issues, which arise when the analysis simultaneously involves multiple modes of information, such as images and text. We intend to implement the proposed methodologies in social sciences and healthcare.