[论文解读] Descriptive vs. inferential community detection in networks: pitfalls, myths, and half-truths

Tiago P. Peixoto|arXiv (Cornell University)|Nov 30, 2021

Complex Network Analysis Techniques被引用 13

一句话总结

本文区分了描述性社区检测与推断性社区检测，指出描述性方法（如模块度最大化）在用于推断性目标时因统计缺陷而失效。文章主张采用基于生成模型（如随机块模型）的推断性方法，以提供统计上合理且稳健的网络结构与形成机制洞察。

ABSTRACT

Community detection is one of the most important methodological fields of network science, and one which has attracted a significant amount of attention over the past decades. This area deals with the automated division of a network into fundamental building blocks, with the objective of providing a summary of its large-scale structure. Despite its importance and widespread adoption, there is a noticeable gap between what is arguably the state-of-the-art and the methods that are actually used in practice in a variety of fields. Here we attempt to address this discrepancy by dividing existing methods according to whether they have a "descriptive" or an "inferential" goal. While descriptive methods find patterns in networks based on context-dependent notions of community structure, inferential methods articulate generative models, and attempt to fit them to data. In this way, they are able to provide insights into the mechanisms of network formation, and separate structure from randomness in a manner supported by statistical evidence. We review how employing descriptive methods with inferential aims is riddled with pitfalls and misleading answers, and thus should be in general avoided. We argue that inferential methods are more typically aligned with clearer scientific questions, yield more robust results, and should be in many cases preferred. We attempt to dispel some myths and half-truths often believed when community detection is employed in practice, in an effort to improve both the use of such methods as well as the interpretation of their results.

研究动机与目标

• 阐明描述性与推断性社区检测方法之间的根本区别。
• 揭示广泛使用的描述性方法（尤其是模块度最大化）中存在的统计缺陷与误解。
• 论证基于生成模型的推断性方法更符合科学探究，能产生更可靠的结果。
• 破除持续导致描述性方法误用的常见误解与半真半假的说法。
• 推动采用有原则的推断性方法，以实现准确、统计上可靠的社区检测。

提出的方法

• 将社区检测方法分类为描述性（基于上下文依赖定义的模式发现）与推断性（基于生成模型的模型拟合）。
• 采用试金石测试：若目标是解释网络形成过程或区分结构与随机性，则必须使用推断性方法。
• 通过随机块模型（SBMs）进行统计推断，将生成模型拟合到数据，利用贝叶斯证据、MDL或BIC实现模型选择。
• 证明推断性方法通过显式建模随机性与模型不确定性，可避免过拟合与分辨率极限问题。
• 使用马尔可夫链蒙特卡洛（MCMC）及合并-分裂MCMC进行社区划分的后验推断。
• 将推断性方法与启发式方法（如模块度最大化、谱聚类、共识聚类）进行对比，突出其统计上的不足。

实验结果

研究问题

• RQ1 描述性与推断性社区检测有何区别？为何这一区分对科学有效性至关重要？
• RQ2 为何尽管广泛应用，模块度最大化在用于推断目的时在根本上存在缺陷？
• RQ3 诸如分辨率参数或显著性检验等常见补救措施，在多大程度上能真正解决过拟合与假阳性问题？
• RQ4 为何诸如“共识聚类可消除过拟合”或“模块度与SBM等价”等误解如此普遍且具有误导性？
• RQ5 推断性方法能否在大规模场景下实际应用？它们是否真如人们所认为的那样更昂贵或效率更低？

主要发现

• 描述性方法（如模块度最大化）在用于推断性目标时在统计上无效，因其缺乏原则性的模型拟合框架。
• 模块度最大化存在分辨率极限与过拟合问题，且任何参数调优或零模型替换都无法完全解决这些问题。
• 共识聚类无法消除过拟合；它仅是在多次运行中聚合了相同的缺陷模式。
• 对质量函数进行的统计显著性检验无法弥补缺乏恰当生成模型的缺陷，因此具有误导性。
• 基于随机块模型的推断性方法可提供统计上有效的推断，能有效分离信号与噪声，并通过贝叶斯证据或信息准则实现模型选择。
• 认为生成模型必须为“真实”才具用处是一种误解——通过MDL、BIC或AIC进行的模型选择可确保即使模型近似，结果依然稳健。