Medical big data classification method and system based on a generative adversarial network and semi-supervised learning

Abstract

The invention discloses a medical big data classification method and system based on a generative adversarial network and semi-supervised learning, and the system comprises a data collection module which is used for collecting medical big data, and obtaining a large amount of medical data and medical images with high data dimension and high class mark uncertainty; The data processing module is used for preprocessing the acquired medical data and medical images; The algorithm application module is used for initializing and training the sub-learners, marking the unlabeled medical data and the unlabeled medical images, and expanding the labeled medical data and the labeled medical images; And the auxiliary decision module is used for classifying the medical big data of the test set. The dataprocessing module further comprises a medical data dimension reduction module, an image processing module, a data classification module and a medical data processing module; The algorithm applicationmodule further comprises a training sample generation module, a training module, a marking module, an expansion module and an integration module. And the accuracy of medical big data classification isimproved.

Description

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine

Login to view more

AI Technical Summary

Benefits of technology

This patented technology uses generational adversarial networks (GNN) or semisupervisory models to classify medical datasets accurately without requiring too much annotated data from both sources. By optimizing an algorithm called gamma net model, this technique helps identify patterns between different types of data distributions more efficiently than traditional methods like binary search techniques alone. Overall, these technical improvements help make medicine work smoother faster and easier compared to manual labelling processes.

Problems solved by technology

This patents describes challengings faced during developing effective computerized techniques for diagnosing diseases like cancer due to limited availabilities of annotated examples. Current approaches involve either relying solely upon randomly distributed labels or heavily weighted versions of these types of approximations. These limitations make them impracitive at identifying rare cases where none exist but they require extensive exploration before finding any relevant ones. Therefore, new ways must be developed to increase the likelihood of successful identification of specific disease patterns even if few patients actually meet those requirements.

Images