Differential privacy protection deep learning algorithm for adaptively allocating dynamic privacy budget

Abstract

The invention aims to provide a differential privacy protection deep learning algorithm for adaptively allocating dynamic privacy budget, which comprises the steps of firstly, giving a data set, setting and initializing a neural network NN, and training the neural network NN by using the data set to obtain a deep learning model M without privacy protection; according to the trained deep learning model M without privacy protection, calculating the average feature correlation by using an LRP algorithm; calculating a correlation ratio; and finally, re-initializing the neural network NN, setting the number of iterations of training, and adding noise in the training process according to the correlation ratio to obtain a deep learning model DPM with differential privacy protection, thereby being capable of protecting the data privacy when the model is used for prediction.

Description

technical field [0001] The invention belongs to the technical field of information security, and in particular relates to a differential privacy protection deep learning algorithm for adaptively allocating a dynamic privacy budget. Background technique [0002] With the development of Internet technology, hundreds of millions of data are generated every day in daily life, and these massive data often contain potential, regular, and ultimately understandable knowledge or patterns. Data mining (DataMining, DM) technology can discover and extract these useful information from these massive data, and give feedback to guide business and human life. It mainly uses machine learning methods and statistical knowledge principles for knowledge mining. Machine learning methods Research and improvement often have an important impact on the efficiency and results of data mining. Deep learning (Deep Learning) is a branch of machine learning, which is an algorithm that attempts to perform ...

AI Technical Summary

Problems solved by technology

[0005] The purpose of the present invention is to provide a differential privacy protection deep learning algorithm for adaptively allocating dynamic privacy budgets, which solves the problem in the prior art that the consumption of privacy budgets is too large, resulting in a low level of privacy protection

Images