Problems of Privacy and Heterogeneity for Federated Learning Applications in Medical Image Analysis
Year: 2024
Type: article
Abstract: Recently, machine learning has become one of \nthe most promising directions in working with medical data. \nDeep neural network models are the most eff ective and \naccurate, but they require large volumes of information for \ntraining. This is a common problem in the case of medical \ndata, especially images, as their creation involves signifi cant \ncosts. \nOne solution to improve the quality of deep neural \nnetwork models without increasing the training dataset is \nmodel aggregation. However, a problem arises with preserving the confi dentiality of medical images. For example, if \none model is trained on an image containing information \nabout a specifi c patient, other models participating in the \naggregation may also gain access to this information. As a \nresult, information about a specifi c patient may be disclosed. \nIn an attempt to address the problem described above, \nthis work aims to research and develop methods for \naggregating machine learning models while preserving the \nprivacy of medical images, particularly federated learning \nmethods
Authors: Himbitski, A., Zelenkovsky, V., Zhydovich, M., Kovalev, V.
Language: English
Subfield: Artificial Intelligence
Field: Computer Science
Domain: Physical Sciences
Open Access status: green
OpenAlex ID: W7042646860