PEnBayes: A Multi-Layered Ensemble Approach for Learning Bayesian Network Structure from Big Data

Abstract

Discovering the Bayesian network (BN) structure from big datasets containing rich causal relationships is becoming increasingly valuable for modeling and reasoning under uncertainties in many areas with big data gathered from sensors due to high volume and fast veracity. Most of the current BN structure learning algorithms have shortcomings facing big data. First, learning a BN structure from the entire big dataset is an expensive task which often ends in failure due to memory constraints. Second, it is quite difficult to select a learner from numerous BN structure learning algorithms to consistently achieve good learning accuracy. Lastly, there is a lack of an intelligent method that merges separately learned BN structures into a well structured BN network. To address these shortcomings, we introduce a novel parallel learning approach called PEnBayes (Parallel Ensemble-based Bayesian network learning). PEnBayes starts with an adaptive data preprocessing phase that calculates the Appropriate Learning Size and intelligently divides a big dataset for fast distributed local structure learning. Then, PEnBayes learns a collection of local BN Structures in parallel using a two-layered weighted adjacent matrix-based structure ensemble method. Lastly, PEnBayes merges the local BN Structures into a global network structure using the structure ensemble method at the global layer. For the experiment, we generate big data sets by simulating sensor data from patient monitoring, transportation, and disease diagnosis domains. The Experimental results show that PEnBayes achieves a significantly improved execution performance with more consistent and stable results compared with three baseline learning algorithms.

Scalable and Hybrid Ensemble-Based Causality Discovery

“…To achieve probabilistic graphical model ensemble, using the three categories explained in Section II, existing ensemble learning approaches can also be categorized into 1) algorithm ensemble for work at [17], 2) data ensemble work at [14], [30], and 3) hybrid ensemble for both data and algorithm at [28] and [7]. In algorithm ensemble category, [17] supports parallel ensemble learning of multiple classifiers on the same data.…”

“…[30] is also a data ensemble approach for network learning from big datasets to achieve better scalability and accuracy. As a hybrid ensemble approach, [28] conducts two-phase (algorithm ensemble for each data partition and data ensemble for multiple data partitions) Bayesian network ensemble learning. The main differences of this work and [28] are: 1) this work first conducts data ensemble among all data partitions and then algorithm ensemble for different algorithms where [28] first conducts algorithm ensemble then data ensemble; 2) our algorithm-level ensemble belongs to heterogeneous ensemble because each learning algorithm uses its own causality discovery models, while [28] belongs to homogeneous ensemble with different learning algorithms of the same Bayesian network model; 3) this paper targets causality discovery instead of Bayesian network learning.…”

Distributed Learning of Generalized Linear Causal Networks

“…The probability density of Distributed causal graph learning information to form a global causal graph that satisfies the acyclicity constraint. One approach is to iterate over local datasets (once) and then combine the local graphs or the local p-values to form a global graph (Na and Yang 2010;Gou et al 2007;Tang et al 2019). However, there is no theoretical guarantee that aggregating local estimates, using this single-iteration approach, would lead to an estimate close to the global minimizer of the loss on the combined data.…”

Algoritmo De Clasificación Mediante Un Enfoque De Machine Learning Y Su Aplicación Al Estudio Meteorológico

“…Big Data es una tecnología que se ha desarrollado debido exponenciales crecimientos en volumen de datos a nivel mundial y las limitaciones de los microprocesadores actuales. Están técnicas habilitan el cómputo para trabajar una tarea compleja en múltiples nodos, con varios núcleos y memoria agrupada (Franke et al, 2016;Qaffas et al,2021;Tang et al,2019;van Evert et al,2017). Por su parte, IoT recolecta la información, es un conjunto de componentes electrónicos que tienen la capacidad de conectarse y comunicar datos (Qaffas, et al, 2021).…”