Adversarial generative network for defending text malicious sample and training method thereof

Abstract

The invention discloses an adversarial generative network for defending a text malicious sample and a training method thereof. A Generator and a Discriminator in an adversarial generative network framework are used for defending and generating the malicious sample. The generator part is composed of an auto-encoder, discrete text data is mapped into a continuous high-dimensional hidden space, and therefore the generator can generate malicious text through a hidden vector. The discriminator is a discrimination model and is used for identifying data. A malicious text generated by the generator ismarked with a real label and input into the discrimination model together with a real sample, so as to train the discrimination model. By adding the discrimination model trained by malicious samples,text data can be identified accurately and efficiently. The generator trains the evaluation score of the malicious sample and the difference between the text data and the malicious sample by using adiscrimination model to generate a malicious sample with stronger attack force. Due to the addition of malicious samples in the training process and the network training process of resistance, the text data network recognition capability, the anti-interference capability and the defense capability are greatly improved.

Description

technical field [0001] The invention belongs to the technical field of text data processing, and more specifically relates to an adversarial generation network and a training method thereof for defending against malicious text samples. Background technique [0002] Malicious samples have been discovered in image recognition and text processing in recent years, and they are extremely offensive to machine learning and deep learning in the field of text data processing. Malicious samples are adversarial samples. Adversarial samples add disturbances that are imperceptible to the human eye to the data, making the model's label prediction of the data confusing and wrong. Adversarial examples are a major hurdle that various machine learning systems need to overcome. The existence of adversarial examples indicates that the model tends to rely on unreliable features to maximize performance. If the features are perturbed, it will cause the model to misclassify, which may lead to cata...

AI Technical Summary

Problems solved by technology

[0005] Aiming at the above defects or improvement needs of the prior art, the present invention provides an adversarial generation network and its training method for defending mal

Images