Identifying Adversarial Sentences by Analyzing Text Complexity

12/19/2019
by   Hoang-Quoc Nguyen-Son, et al.
0

Attackers create adversarial text to deceive both human perception and the current AI systems to perform malicious purposes such as spam product reviews and fake political posts. We investigate the difference between the adversarial and the original text to prevent the risk. We prove that the text written by a human is more coherent and fluent. Moreover, the human can express the idea through the flexible text with modern words while a machine focuses on optimizing the generated text by the simple and common words. We also suggest a method to identify the adversarial text by extracting the features related to our findings. The proposed method achieves high performance with 82.0 accuracy and 18.4 methods whose the best accuracy is 77.0

READ FULL TEXT

page 1

page 2

page 3

page 4

research
10/15/2019

Detecting Machine-Translated Text using Back Translation

Machine-translated text plays a crucial role in the communication of peo...
research
04/24/2019

Detecting Machine-Translated Paragraphs by Matching Similar Words

Machine-translated text plays an important role in modern life by smooth...
research
05/21/2023

GPT Paternity Test: GPT Generated Text Detection with GPT Genetic Inheritance

Large Language Models (LLMs) can generate texts that carry the risk of v...
research
01/30/2023

ChatGPT or Human? Detect and Explain. Explaining Decisions of Machine Learning Model for Detecting Short ChatGPT-generated Text

ChatGPT has the ability to generate grammatically flawless and seemingly...
research
05/09/2023

Estimating related words computationally using language model from the Mahabharata – an Indian epic

'Mahabharata' is the most popular among many Indian pieces of literature...
research
07/10/2017

Towards Crafting Text Adversarial Samples

Adversarial samples are strategically modified samples, which are crafte...
research
07/23/2023

Towards Automatic Boundary Detection for Human-AI Hybrid Essay in Education

Human-AI collaborative writing has been greatly facilitated with the hel...

Please sign up or login with your details

Forgot password? Click here to reset