arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2012.03540cs.LGcs.AIcs.DB
收藏

贝叶斯网络的高效可扩展结构学习:算法与应用

Efficient and Scalable Structure Learning for Bayesian Networks: Algorithms and Applications

Alibaba Group(阿里巴巴集团);Oxford University(牛津大学);Peking University(北京大学)

Rong Zhu, Andreas Pfadler, Ziniu Wu, Yuxing Han, Xiaoke Yang, Feng Ye, Zhenping Qian, Jingren Zhou, Bin Cui

摘要:

贝叶斯网络的结构学习是一个重要的研究课题。它在阿里巴巴集团的各种应用程序中发挥着核心作用。然而,现有的结构学习算法在实际应用中受到很大的限制,因为它们的效率低,可扩展性差。为了解决这个问题,我们提出了一种新的结构学习算法LEAST,它全面满足我们的业务需求,因为它达到了高精度,效率和可扩展性在同一时间。LEAST的核心思想是将结构学习转化为一个连续的约束优化问题,并使用一个新的可微约束函数来度量结果图的非循环性。与现有的工作不同,我们的约束函数是建立在图的谱半径,可以在近线性时间w.r.t.图节点大小。在此基础上,LEAST算法可以以较低的存储开销高效地实现。根据我们的基准评估,LEAST比最先进的方法快1到2个数量级,具有相当的精度,并且能够在具有数十万个变量的BN上进行扩展。在我们的生产环境中,LEAST部署并服务于20多个应用程序,每天执行数千次。我们描述了一个具体的情况下,在阿里巴巴的机票预订服务,LEAST被应用于建立一个近实时的自动异常检测和根本错误原因分析系统。我们还表明,LEAST解锁BN结构学习应用在新领域的可能性,如大规模基因表达数据分析和可解释的推荐系统。

英文摘要:

Structure Learning for Bayesian network (BN) is an important problem with extensive research. It plays central roles in a wide variety of applications in Alibaba Group. However, existing structure learning algorithms suffer from considerable limitations in real world applications due to their low efficiency and poor scalability. To resolve this, we propose a new structure learning algorithm LEAST, which comprehensively fulfills our business requirements as it attains high accuracy, efficiency and scalability at the same time. The core idea of LEAST is to formulate the structure learning into a continuous constrained optimization problem, with a novel differentiable constraint function measuring the acyclicity of the resulting graph. Unlike with existing work, our constraint function is built on the spectral radius of the graph and could be evaluated in near linear time w.r.t. the graph node size. Based on it, LEAST can be efficiently implemented with low storage overhead. According to our benchmark evaluation, LEAST runs 1 to 2 orders of magnitude faster than state of the art method with comparable accuracy, and it is able to scale on BNs with up to hundreds of thousands of variables. In our production environment, LEAST is deployed and serves for more than 20 applications with thousands of executions per day. We describe a concrete scenario in a ticket booking service in Alibaba, where LEAST is applied to build a near real-time automatic anomaly detection and root error cause analysis system. We also show that LEAST unlocks the possibility of applying BN structure learning in new areas, such as large-scale gene expression data analysis and explainable recommendation system.

机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。

↑