Measuring Conflict Among Objective Functions in Multi-Objective Optimization
Kullman, Nicholas David
MetadataShow full item record
We present a process to measure the conflict among objective functions within and across multi-objective systems. To do so, we introduce a new metric to quantify pairwise objective conflict. We also demonstrate new applications for two existing measures of conflict from evolutionary multi-objective optimization (EMO). To demonstrate this quantification of conflict and the utility of our proposed pairwise objective conflict measure we perform a case study of the impact of climate change on the joint provision of ecosystem services in the Deschutes National Forest. For three climate scenarios, we quantify the total amount of conflict in the multi-objective system. We also compare the conflict across climate scenarios. We find that climate change impacts both the individual and joint provision of ecosystem services and that system conflict increases with increasing climate change severity. The case study demonstrates that our proposed process and new conflict metric successfully quantify and differentiate the amount of conflict within and across multi-objective systems and that they stand to serve as a useful tool for multi-objective decision making.