Parallelise RandomForest and constituent DecisionTree, DataFrame, LossFunction etc. with hybrid OpenMP+MPI parallel processing.
Seems most appropriate for grid searching or a Genetic algorithm such that separate nodes run MPI-parallelised training of specific hyperparameter configurations, then the results are all collated and compared.
Parallelise
RandomForest
and constituentDecisionTree
,DataFrame
,LossFunction
etc. with hybrid OpenMP+MPI parallel processing.Seems most appropriate for grid searching or a Genetic algorithm such that separate nodes run MPI-parallelised training of specific hyperparameter configurations, then the results are all collated and compared.