Missing data filling generation method based on dual-condition generative adversarial network

Abstract

The invention relates to a missing data filling generation method based on a dual-condition generative adversarial network, which belongs to the field of data perception and reconstruction in a computer, and comprises the following steps: S1, encoding sample data, and designing a sample category serving as a data generation condition and an expression mode of existing data of a sample in a generation process; s2, constructing a structure of a dual-condition generative adversarial network, wherein the structure comprises a generative model and a discrimination model; s3, representing a target optimization function of the dual-condition generative adversarial network structure; s4, establishing a training data set of the data generation model, and training the dual-condition generative adversarial network; and S5, analyzing different data missing conditions, and performing missing data generation filling by adopting the trained dual-condition generative adversarial network. The invention provides a method for constructing a high-quality training data set for table data. The method is used for supporting machine learning model training in a big data application scene.

Description

technical field [0001] The invention relates to a missing data filling and generating method based on a dual conditional generation confrontation network, and belongs to the field of data perception and reconstruction in computers. Background technique [0002] In recent years, in machine learning, generative adversarial network models have become increasingly important and popular due to their applicability in different domains. Their ability to represent complex, high-dimensional data can be used to process images, videos, tabular data, and other academic fields. In fields such as digital finance, the generation of tabular data is a core concern of researchers. Given a set of random noises, a generative adversarial network model is able to generate the corresponding tabular data. [0003] In real application scenarios, it is often necessary to output tabular data under given conditions, and tabular data is usually missing. Given the particularity of tabular data, the co...

AI Technical Summary

Problems solved by technology

[0005] However, the above methods still have the following problems: (1) due to the lack of data in the sample data in actual scenarios, and the data imbalance problem of minority class samples; (2) the design only takes category conditions as input, and The quality of the training data set constructed by generating samples cannot be well guaranteed to meet the requirements of high-quality training data sets for machine learning model training in big data application scenarios

Images