Good Seeds Bear Good Fruit: Using Benefit-to-Cost Ratios in Multiobjective Spatial Optimization under Epistasis
研究了在存在上位性（相互依赖）的情况下，使用效益成本比排名作为进化算法起点进行多目标空间优化的效果，发现该方法在决策空间表现良好，但目标空间效果需模型评估。
Many biophysical models exhibit epistasis (interdependence), where a conservation action impacts the effectiveness of another elsewhere. At the same time, ranking conservation actions according to the independent benefit-to-cost ratios is cost-efficient when epistasis is absent. We use benefit-to-cost rankings as starting points for an evolutionary algorithm employing an epistatic biophysical model. We model a variety of conservation actions to assess trade-offs for sediment reduction and wildlife conservation in the study watershed. We find that despite the presence of epistasis, the weighted benefit-to-cost ratio-derived solutions perform remarkably well in the decision space, but effects in objective space need the model evaluation. <i></i>