公开(公告)号：US10360517B2

公开(公告)日：2019-07-23

申请号：US15822462

申请日：2017-11-27

Applicant: SAS Institute Inc.

Inventor： Patrick Nathan Koch ,  Brett Alan Wujek ,  Oleg Borisovich Golovidov ,  Steven Joseph Gardner ,  Joshua David Griffin ,  Scott Russell Pope ,  Yan Xu

IPC: G06N20/00 ,  G06F16/2457 ,  G06N3/08 ,  G06N3/12 ,  G06N5/00 ,  G06N20/10

Abstract: A computing device automatically selects hyperparameter values based on objective criteria to train a predictive model. Each session of a plurality of sessions executes training and scoring of a model type using an input dataset in parallel with other sessions of the plurality of sessions. Unique hyperparameter configurations are determined using a search method and assigned to each session. For each session of the plurality of sessions, training of a model of the model type is requested using a training dataset and the assigned hyperparameter configuration, scoring of the trained model using a validation dataset and the assigned hyperparameter configuration is requested to compute an objective function value, and the received objective function value and the assigned hyperparameter configuration are stored. A best hyperparameter configuration is identified based on an extreme value of the stored objective function values.