Comparative study for inference of hidden classes in stochastic block models
Comparative study for inference of hidden classes in stochastic block models
复制标题
DOI：
10.1088/1742-5468/2012/12/p12021
复制
发表时间：
2012-12-01
影响因子：
2.4
通讯作者：
Zdeborova, Lenka
中科院分区：
文献类型：
--
作者：
Zhang, Pan;Krzakala, Florent;Zdeborova, Lenka
随机块模型中隐藏类的推断是一个具有重要应用的经典问题。解决此问题最常用的方法包括朴素平均场方法或启发式谱方法。最近，针对这个问题提出了信念传播。在这篇文章中，我们对综合创建的网络的三种方法进行了比较研究。我们表明，与朴素平均场和谱方法相比，置信传播表现出更好的性能。这适用于准确性、计算效率和过度拟合数据的趋势。
Inference of hidden classes in stochastic block models is a classical problem with important applications. Most commonly used methods for this problem involve naive mean field approaches or heuristic spectral methods. Recently, belief propagation was proposed for this problem. In this contribution we perform a comparative study between the three methods on synthetically created networks. We show that belief propagation shows much better performance when compared to naive mean field and spectral approaches. This applies to accuracy, computational efficiency and the tendency to overfit the data.