Dynamic community detection has been prospered as a powerful tool for quantifying changes in dynamic brain network connectivity patterns by identifying strongly connected sets of nodes. However, as the network science problems and network data to be processed become gradually more sophisticated, it awaits a better method to efficiently learn low dimensional representation from dynamic network data and reveal its latent function that changes over time in the brain network. In this work, an adversarial temporal graph representation learning (ATGRL) framework is proposed to detect dynamic communities from a small sample of brain network data. It adopts a novel temporal graph attention network as an encoder to capture more efficient spatio-temporal features by attention mechanism in both spatial and temporal dimensions. In addition, the framework employs adversarial training to guide the learning of temporal graph representation and optimize the measurable modularity loss to maximize the modularity of community. Experiments on the real-world brain networks datasets are demonstrated to show the effectiveness of this new method.

本文提出了一种新的方法——对抗时变图表示学习(ATGRL)，用于从大脑网络的少量样本数据中检测动态社区。该方法采用了一种新型的时变图注意力网络作为编码器，通过空间和时间维度上的注意力机制来捕获更高效的时空特征，并采用对抗性训练来指导时变图表示的学习，并最大化社区的模块度度量损失，实验证明了该方法的有效性。

通过对抗时空图表示学习进行动态社区检测