TextFirewall: Omni-Defending Against Adversarial Texts in Sentiment Classification
Sign Up to like & getrecommendations! Published in 2021 at "IEEE Access"
DOI: 10.1109/access.2021.3058278
Abstract: Sentiment classification has been broadly applied in real life, such as product recommendation and opinion-oriented analysis. Unfortunately, the widely employed sentiment classification systems based on deep neural networks (DNNs) are susceptible to adversarial attacks with… read more here.
Keywords: sentiment classification; texts; adversarial texts; defending adversarial ... See more keywords