Open access
Jul 2026
Prompt-Driven Fuzzing Debiasing Framework for Robust Visual Question Answering
A unified prompt-driven debiasing framework that integrates generative prompt learning and a fuzzing-based bias correction mechanism is proposed, which significantly improves both in-distribution accuracy and out-of-distribution robustness, outperforming existing prompt-only or data-augmentation-only debiasing methods.
Ya-Li Fan, Gang-Yu Huang, Qiwen Lu et al.
· Multimodal Technologies and... · 0 citations