Pre-training language model-oriented privacy disclosure risk assessment method and system

Abstract

The invention relates to the field of privacy security, and aims to provide a pre-training language model-oriented privacy disclosure risk assessment method and system. Comprising the following steps: adding forged data into a pre-training data set; inputting the pre-training data set into the initialized neural network model, and calculating loss according to a set pre-training task and a loss function; parameters of the model are continuously updated in the training process, and the privacy leakage risk of the model is increased; inputting the fine tuning data set into a pre-trained neural network model, and performing fine tuning on the feature extraction capability of the model; privacy prefix content is input into the model, and text information serving as a prediction result is output; and calculating, counting and sorting the confusion of the output information, and evaluating the risk of privacy data leakage by comparing the proportion of the generated privacy information. According to the method, the accuracy of evaluating the privacy data leakage risk can be effectively improved, the privacy data leakage risk existing in the pre-training language model is exposed, and a thought is provided for subsequent development of related defense methods.

Description

technical field [0001] The invention relates to the technical field of privacy security, and in particular, to a method and system for assessing privacy leakage risk for pre-training language models. Background technique [0002] Natural language processing is an important application and branch in the field of artificial intelligence. Its purpose is to use deep learning and other technologies to intelligently process natural language. In recent years, pre-trained language models based on Transformer structure such as GPT-2 have gradually become one of the mainstream models for natural language processing tasks due to their superior performance. [0003] In the era of big data, in order to obtain pre-trained language models with superior performance, trainers often obtain a large amount of data for training in various ways, such as crawling identity information on social networks or using private information uploaded by user terminals. The data may contain sensitive informa...

AI Technical Summary

Problems solved by technology

[0005] However, the existing research results mainly analyze the privacy leakage risk of the large-scale pre-trained language model in the inference stage, and there are few researches on the possible privacy leakage risk of the model in the pre-training stage.

The lack of work in this area makes the analysis of the risk of privacy leakage of the pre-trained language model have loopholes, leading to increased privacy threats to the deep learning model

Images