Experimental result

Our proposed GDST-PPO outperforms six baseline methods in terms of efficiency, resin utilization, and robustness metrics in scientific workflows of different sizes. For each scenario, data were collected through 10 replicated trials, averaged and subjected to a paired t-test/Wilcoxon signed rank test.

我们提出的GDST-PPO在不同规模的科学工作流中，于效率、树脂利用率以及鲁棒性指标三个维度上均优于六种基线方法。针对每一实验场景，我们通过10次重复试验采集数据，对所得结果取平均值后开展配对t检验/威尔科克森符号秩检验。