Many practical search and optimization problems are ideally posed as multi-criterion optimization problems due to pragmatic difficulties of combining multiple conflicting goals of a problem-solving task into a single one. Theoretically, these problems give rise to a set of Pareto-optimal solutions, which must be discovered by an optimization method and a multi-criterion decision-making (MCDM) approach must be applied a-priori, a-posteriori, or interactively in consultation with human decision-makers (DMs) to select a single preferred solution. Efficient evolutionary multi-objective optimization (EMO) algorithms have been proposed, challenging and systematic test problems and performance metrics are suggested, and real-world applications in societal, engineering, science and business are executed to demonstrate the importance of EMO methods. Since early seventies, several MCDM approaches were proposed, mainly by mathematics and economics researchers, to select the single-most preferred Pareto-optimal solution with a few decision-maker (DM) calls. Both EMO and MCDM tasks are essential for addressing multi-criterion optimization problems, but EMO researchers have mostly ignored the MCDM part mainly due to the dependance on and inaccessibility of human DMs.

This tutorial plans to present a quick overview of the fundamentals, followed by a brief discussion of a few popularly-used EMO algorithms including a brief presentation of commonly-used test problems, performance metrics, and application case studies. A brief outline of a few popular MCDM approaches will also be introduced to the participants. But the main part of the tutorial will be spent on describing a number of advancements in the EMO field, that are collectively revealing its importance in various practical problem-solving tasks, such as, partial Pareto-set search, knowledge discovery, robust- and reliability-based optimization, multi-objectivization techniques, regularized EMO, innovation path discovery, bilevel multi-objective optimization, surrogate-assisted optimization and others. These topics will provide a much fuller understanding of the scope of EMO methods and provide new research ideas to the participants.

Another salient aspect of the tutorial will be to introduce a Machine-DM concept, proposed recently to emulate the behavior of human DMs through trained machine learning (ML) models, leading to, for the first time, a number of benchmark problems for interactive MCDM (iMCDM). With publicly available ML-based machine-DMs for certain benchmark problems, EC/EMO and ML researchers can now focus on developing new and efficient MCDM approaches and comparing with existing MCDM approaches. The Machine-DM concept should be of interest, not only to EC researchers, but also to ML researchers and GECCO-2026 will be a great venue for this tutorial.

Finally, to spur further interests, a hands-on computer simulation of a few EMO algorithms for finding a set of trade-off Pareto solutions and the proposed Machine-DM concept through a number of Bench-iMCDM methods to select a single preferred solution will be demonstrated. Github codes will be introduced and their scope and use will be demonstrated.