Traditionally, learning the structure of a Dynamic Bayesian Network has been centralized, with all data pooled in one location. However, in real-world scenarios, data are often dispersed among multiple parties (e.g., companies, devices) that aim to collaboratively learn a Dynamic Bayesian Network while preserving their data privacy and security. In this study, we introduce a federated learning approach for estimating the structure of a Dynamic Bayesian Network from data distributed horizontally across different parties. We propose a distributed structure learning method that leverages continuous optimization so that only model parameters are exchanged during optimization. Experimental results on synthetic and real datasets reveal that our method outperforms other state-of-the-art techniques, particularly when there are many clients with limited individual sample sizes.

翻译：传统上，动态贝叶斯网络的结构学习通常采用集中式方法，将所有数据汇集于单一位置。然而，在实际应用场景中，数据往往分散在多个参与方（例如企业、设备）之间，这些参与方希望在保护数据隐私与安全的前提下，协作学习动态贝叶斯网络。本研究提出了一种联邦学习方法，用于从水平分布于不同参与方的数据中估计动态贝叶斯网络的结构。我们设计了一种分布式结构学习方法，该方法利用连续优化技术，使得在优化过程中仅需交换模型参数。在合成数据集和真实数据集上的实验结果表明，我们的方法优于其他现有先进技术，尤其在参与方众多且各客户端样本量有限的情况下表现更为突出。