Given a graph with millions of nodes, what patterns exist in the distributions of node characteristics? How can we detect them and separate anomalous nodes in a way similar to human visual perception? More generally, how can we identify micro-clusters in a histogram and spot some interesting patterns?

中文翻译：

---

EagleMine：视觉引导微团簇识别和集体异常检测

给定一个包含数百万个节点的图，节点特征的分布存在什么模式？我们如何以类似于人类视觉感知的方式检测它们并分离异常节点？更一般地说，我们如何识别直方图中的微簇并发现一些有趣的模式？