Machine learning model-oriented data recovery method and related device

Abstract

The invention provides a data recovery method for a machine learning model, and the method can adjust the training process of the machine learning model, so as to enable the model to record training set data through the linear combination of model weights contained in the model. And the training set data can be recovered through the model weight value after linear combination training under the condition that the training set data cannot be contacted. As all the model weights are adopted for recording, compared with existing correlation coding attacks, the accuracy is higher, and the performance of the model is not prone to being reduced. The invention further provides a machine learning model-oriented data recovery device, electronic equipment and a storage medium, which have the above beneficial effects.

Description

technical field [0001] The invention relates to the field of machine learning, in particular to a data recovery method, device, electronic device and storage medium oriented to a machine learning model. Background technique [0002] In the field of machine learning, the trained machine learning model is usually unable to directly restore the training set. However, if the training method of the machine learning model is modified, and the model is guided to actively record the training set data by using the model parameters during the training process, the training set can be restored by using the model parameters after training. Correlation value coding attack is a training method that can guide the machine learning model to record the training set. It uses the Pearson correlation coefficient to maximize the correlation between the model parameters and the data to be recorded to complete the data recording. However, the correlation value coding attack has some problems, such...

AI Technical Summary

Problems solved by technology

However, there are some problems in the correlation value encoding attack, such as limited recording ability, the recovered data has a large error compared with the original data, and it will also reduce the accuracy of the test set of the model.

Images