METHOD AND SYSTEM FOR GENERATING SYNTHETIC DATA Russian patent published in 2024 - IPC G06F16/906 G06N3/08 

FIELD: data processing.

SUBSTANCE: invention relates to a method and a system for generating synthetic data. Method comprises obtaining an original data sample containing information on data subjects, and at least one set of data associated with said data subjects; classifying information contained in the data sample to identify attributes related to sensitive data (SD) and attributes not related to SD; at least one synthetic data generator is trained using machine learning algorithms based on: data sample; values of SD classes obtained at the previous stage; metadata about data for training and primary and foreign keys, linking information about data subjects and data set; at least one set of synthetic data is generated by means of a trained generator, in which the set of attributes of the data subjects differs from the sets of attributes of the data subjects contained in the original data sample, with preservation of primary and foreign keys containing information on the data subjects.

EFFECT: possibility of generating synthetic data, the structure of which corresponds to the original data, while ensuring confidentiality of sensitive data.

10 cl, 7 dwg, 10 tbl