Single MCMC chain parallelisation on decision trees

Abstract Decision trees (DT) are highly famous in machine learning and usually acquire state-of-the-art performance. Despite that, well-known variants like CART, ID3, random forest, boosted miss a probabilistic version that encodes prior assumptions about tree structures shares statistical strength between node parameters. Existing work on Bayesian DT depends Markov Chain Monte Carlo (MCMC), which can be computationally slow, especially high dimensional data expensive proposals. In this study, we propose method to parallelise single MCMC chain an average laptop or personal computer enables us reduce its run-time through multi-core processing while the results statistically identical conventional sequential implementation. We also calculate theoretical practical reduction run time, obtained utilising our multi-processor architectures. Experiments showed could achieve 18 times faster running time provided serial parallel implementation identical.