Federal learning system for multi-type task image analysis

Abstract

The invention belongs to the technical field of images, and discloses a federated learning system for multi-type task image analysis, the federated learning system comprises a Server and a Client, the Client comprises a Decoder, an Encoder, an Encoder average value and fc, and the Server comprises the Encoder average value. According to the method, data islands among all parties are broken through by using a federated learning framework, so that modeling can be carried out by combining data of all parties on the premise of protecting image data privacy, the accuracy is higher than that of independent modeling, and image analysis of multiple types of tasks can be carried out at the same time.

Description

technical field [0001] The invention belongs to the field of image technology, in particular to a federated learning system for multi-type task image analysis, and is especially suitable for medical treatment. Background technique [0002] The medical image analysis system is a technology that uses computers to automatically process, feature extract and classify medical images. The main objects of analysis are human cell smear images, X-ray photographs and ultrasound images of various parts of the human body. In recent years, with the rapid development of big data, deep learning, and artificial intelligence technologies, some analysis methods based on deep learning and artificial intelligence have also appeared in the field of medical images. [0003] In an existing medical image classification or segmentation method, one party's data is directly used for training, but due to the lack of data, the accuracy of the model trained by this method is low; in another medical image ...

AI Technical Summary

Problems solved by technology

This method does protect data privacy, but it requires that the tasks of all parties must be the same, that is, either classification or segmentation, so this method cannot be used for joint training of multiple different types of tasks, and because The heterogeneous type of data from all parties, such as Non-iid, that is, non-independent and identical distribution, so the global model trained under the federated learning framework is not effective in classifying or segmenting data from all parties, and the accuracy is low

Images