Chengxiang Yuan
Papers
2
Total Citations
9
H-Index
2
About
Chengxiang Yuan is a researcher dedicated to enhancing the robustness and security of natural language processing (NLP) systems, particularly intelligent question-and-answer (Q&A) robots. His work addresses a critical vulnerability: how minor adversarial perturbations, such as a single typo, can cause NLP models to fail. Yuan’s major contributions include developing DPAEG, a dependency parse-based adversarial examples generation method that systematically creates subtle, grammar-aware attacks to expose weaknesses in Q&A robots. This approach, detailed in his 2020 paper (7 citations), provides a more realistic and challenging test for model defenses. His earlier 2019 analysis (2 citations) laid the groundwork by formally assessing the fragility of AI Q&A systems. By pioneering methods to both generate and evaluate adversarial examples, Yuan is helping to build more secure and reliable conversational AI, ensuring that intelligent robots can withstand real-world input errors and malicious manipulation.
Research Focus
Key Achievements
Top Papers
- 1
- 2