Structural Bias in Multi-objective Optimization
This work introduces a suite of synthetic multi-objective test problems with analytically controlled Pareto fronts and deliberately uninformative objective values, designed to decouple algorithmic behaviour from problem structure, allowing bias induced purely by algorithmic operators and design choices to be observed.