2017/04/27 by Stefano Beretta, Mauro Castelli, Beretta, Stefano +8
Computer Science · Decision Sciences · #Artificial Intelligence (cs.AI) #Bayesian Modeling and Causal Inference #Data Quality and Management #FOS: Computer and information sciences #Machine Learning (cs.LG) #Machine Learning (stat.ML) #Machine Learning and Data Classification
paper · pdf · doi:10.48550/arxiv.1704.08676
openalex publication_date 2017/04/27 · openalex created_date 2022/10/06 · openalex updated_date 2026/07/28
One of the most challenging tasks when adopting Bayesian Networks (BNs) is\nthe one of learning their structure from data. This task is complicated by the\nhuge search space of possible solutions, and by the fact that the problem is\nNP-hard. Hence, full enumeration of all the possible solutions is not always\nfeasible and approximations are often required. However, to the best of our\nknowledge, a quantitative analysis of the performance and characteristics of\nthe different heuristics to solve this problem has never been done before.\n For this reason, in this work, we provide a detailed comparison of many\ndifferent state-of-the-arts methods for structural learning on simulated data\nconsidering both BNs with discrete and continuous variables, and with different\nrates of noise in the data. In particular, we investigate the performance of\ndifferent widespread scores and algorithmic approaches proposed for the\ninference and the statistical pitfalls within them.\n