CMA-ES

This paper proposes diversified residual symbolic regression (DRSR), which uses the quality-diversity paradigm to collect accurate symbolic expressions with diverse residual patterns. It enables users to select expressions based on domain knowledge and reveals multiple underlying relationships in synthetic and real-world datasets.

This paper proposes DX-NES-ICI, a natural evolution strategy that accounts for implicit constraints and integer variables in mixed-integer black-box optimization. It demonstrates improved performance over CMA-ES w. Margin on benchmark problems, particularly when continuous variables contribute more to the objective function value than integer variables.