Logistic regression modeling method using secret variance

摘要

Methods, systems, and devices that include computer programs encoded on computer storage media for training multi-party secure logistic regression models (SLRMs) using secret sharing techniques. One method is to use secret sharing (SS) to split the sample training data for a secure logistic regression model (SLRM) into multiple shares, where each share is a secure compute node (SCN). Iteratively updates the parameters associated with the SLRM using the steps and each share of the sample training data distributed to, and the iterative updates continue until predetermined conditions emerge. Includes steps and steps that iteratively update the parameters associated with SLRM and then output training results configured for use by each SCN.