Differential privacy federal learning method for resisting member reasoning attack

Abstract

The invention discloses a differential privacy federal learning method for resisting member reasoning attacks. The method specifically comprises the steps that each client uses local data for training to generate an adversarial generative network model and generate false data; for each round of federated learning communication, the server side randomly selects a client side participating in the round of communication, and issues global network model parameters and a loss function and an optimizer adopted in the training process; the selected client uses the false data to train a global network model and sends parameters of the trained global network model back to the server; the server side adopts a federated average aggregation method to update global network model parameters; and the server side judges whether the next communication is continued or not, if so, the global network model parameters are continued to be published, otherwise, the communication is ended, and the global network model parameters are saved. According to the method, the data privacy of the client is further protected under the condition of original data islands, and member reasoning attacks can be resisted.

Description

technical field [0001] The invention relates to the technical field of machine learning, in particular to a differential privacy federated learning method for resisting member reasoning attacks. Background technique [0002] Federated learning is a distributed machine learning framework with privacy-preserving technology, which aims to assist in the training of machine learning models through decentralized clients without data leakage. Under the federated learning architecture, the problem of collaboration between different data owners without exchanging data is solved by designing fake models. Since data is not transferred, it can effectively protect user privacy or affect data specifications. [0003] However, under resource-constrained conditions such as fewer training data sets, federated learning often performs poorly, and under such conditions, the training of federated learning will be hindered. At the same time, when some clients have insufficient data, the global ...

AI Technical Summary

Problems solved by technology

[0003] However, under resource-constrained conditions such as fewer training data sets, federated learning often performs poorly. Under such conditions, the training of federated learning will be hindered.

At the same time, when some clients have insufficient data, the global network model trained by these clients will have a negative impact on the overall global network model when aggregated on the server side

[0004] In terms of defending against membership inference attacks, the existing methods of federated learning generally add noise to the parameters of the global network model in the process of client training the global network model. Although this method can play a role in resisting membership inference attacks, this method tends to significantly degrade the performance of global network models

This method of adding noise to model parameters during the training process obtains the characteristics of resisting member inference attacks at the expense of larger model performance. The defect is obvious and needs to be improved.

Images