FISA improves MLLM visual question answering by generating and filtering augmented images based on the model's own failure cases.
Smith and Daniel Khashabi and Hannaneh Hajishirzi , title =
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.AI 1years
2026 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
Failure-Informed Image Self-Augmentation for Multimodal Large Language Model Self-Improvement
FISA improves MLLM visual question answering by generating and filtering augmented images based on the model's own failure cases.