Declaration-based Prompt Tuning for Visual Question Answering
This addresses a bottleneck in adapting pre-trained visual-language models to downstream tasks like VQA, reducing the need for labeled data and improving generalization, though it is incremental as it builds on existing prompt-tuning paradigms.
The paper tackles the inconsistency between pre-training and fine-tuning objectives in visual question answering (VQA), proposing Declaration-based Prompt Tuning (DPT) to jointly optimize both, resulting in accuracy improvements of 2.68% in fully-supervised and over 31% in zero-shot/few-shot settings on the GQA dataset.
In recent years, the pre-training-then-fine-tuning paradigm has yielded immense success on a wide spectrum of cross-modal tasks, such as visual question answering (VQA), in which a visual-language (VL) model is first optimized via self-supervised task objectives, e.g., masked language modeling (MLM) and image-text matching (ITM), and then fine-tuned to adapt to downstream task (e.g., VQA) via a brand-new objective function, e.g., answer prediction. The inconsistency of the objective forms not only severely limits the generalization of pre-trained VL models to downstream tasks, but also requires a large amount of labeled data for fine-tuning. To alleviate the problem, we propose an innovative VL fine-tuning paradigm (named Declaration-based Prompt Tuning, abbreviated as DPT), which jointly optimizes the objectives of pre-training and fine-tuning of VQA model, boosting the effective adaptation of pre-trained VL models to the downstream task. Specifically, DPT reformulates the objective form of VQA task via (1) textual adaptation, which converts the given questions into declarative sentence-form for prompt-tuning, and (2) task adaptation, which optimizes the objective function of VQA problem in the manner of pre-training phase. Experimental results on GQA dataset show that DPT outperforms the fine-tuned counterpart by a large margin regarding accuracy in both fully-supervised (2.68%) and zero-shot/few-shot (over 31%) settings. All the data and codes will be available to facilitate future research.