Qiyin Dai
Papers
1
Total Citations
6
H-Index
1
About
Qiyin Dai is a researcher specializing in the security and reliability of deep learning systems, with a particular focus on automated testing and fuzzing methodologies. Their most notable contribution is the development of CAGFuzz (Coverage-Guided Adversarial Generative Fuzzing), a pioneering framework introduced in their 2019 paper that addresses critical vulnerabilities in Deep Neural Networks (DNNs). This work tackles the pressing challenge of DNNs' tendency to produce erroneous outputs due to limited training datasets and reliance on manual labeling—issues that have real-world implications for safety-critical applications like unmanned vehicles, speech processing, and robotics. By combining coverage-guided testing with adversarial generative techniques, Dai's approach systematically identifies and exposes weaknesses in deep learning models, advancing the field of AI robustness testing. With their work accumulating citations and influencing subsequent research, Dai stands at the intersection of software testing and artificial intelligence, contributing essential tools for ensuring the dependability of increasingly autonomous systems. Their research continues to shape how developers and researchers validate deep learning models against unforeseen inputs and edge cases.
Research Focus
Key Achievements
Top Papers
- 1