2016/10/02 by Joseph Ramsey, Ramsey, Joseph · 3 citations
Computer Science · Decision Sciences · #Advanced Bandit Algorithms Research #Artificial Intelligence (cs.AI) #Bayesian Modeling and Causal Inference #FOS: Computer and information sciences #Machine Learning and Algorithms #Machine Learning and Data Classification
paper · pdf · doi:10.48550/arxiv.1610.00378
openalex publication_date 2016/10/02 · openalex created_date 2025/10/10 · openalex updated_date 2026/07/28
A number of attempts have been made to improve accuracy and/or scalability of\nthe PC (Peter and Clark) algorithm, some well known (Buhlmann, et al., 2010;\nKalisch and Buhlmann, 2007; 2008; Zhang, 2012, to give some examples). We add\nhere one more tool to the toolbox: the simple observation that if one is forced\nto choose between a variety of possible conditioning sets for a pair of\nvariables, one should choose the one with the highest p-value. One can use the\nCPC (Conservative PC, Ramsey et al., 2012) algorithm as a guide to possible\nsepsets for a pair of variables. However, whereas CPC uses a voting rule to\nclassify colliders versus noncolliders, our proposed algorithm, PC-Max, picks\nthe conditioning set with the highest p-value, so that there are no\nambiguities. We combine this with two other optimizations: (a) avoiding\nbidirected edges in the orientation of colliders, and (b) parallelization. For\n(b) we borrow ideas from the PC-Stable algorithm (Colombo and Maathuis, 2014).\nThe result is an algorithm that scales quite well both in terms of accuracy and\ntime, with no risk of bidirected edges.\n