谢谢审稿人的宝贵意见。在本文的实验中，我们分析了 范围内的总体精度，并

谢谢审稿人的宝贵意见。在本文的实验中，我们分析了 范围内的总体精度，并将设置为0.9，如下图1所示。当取0.9时，在Indian、Salinas和WHU-Hi-HongHu这三个数据集上均取得最好的精度。因为随着逐渐减小，模型中包含的重要图结构信息越少，影响了模型的分类性能。因此，对于这三个数据集，在实验中均将设置为0.9。

Thank you for your valuable feedback from the reviewers. In this experiment, we analyzed the overall accuracy within the range and set it to 0.9, as shown in Figure 1 below. When taken as 0.9, the best accuracy was achieved on the Indian, Salinas, and WHU-Hi HongHu datasets. Because as it gradually decreases, the important graph structure information contained in the model decreases, which affects the classification performance of the model. Therefore, for all three datasets, they will be set to 0.9 in the experiment.

Thank you for your valuable comments. In this experiment, we analyzed the overall accuracy in the range and set it to 0.9, as shown in Figure 1 below. When 0.9 is taken, the best accuracy is achieved on Indian, Salinas and WHU-Hi-HongHu data sets. Because with the gradual decrease, the model contains less important graph structure information, which affects the classification performance of the model. Therefore, for these three data sets, it will be set to 0.9 in the experiment.