Zero-shot Visual Question Answering with Language Model Feedback

TL;DR

This paper introduces LAMOC, a novel approach that improves zero-shot knowledge-based visual question answering by using language model feedback to guide captioning, achieving competitive results without fine-tuning.

Contribution

The paper presents a new language model guided captioning method that leverages feedback from a PLM to enhance VQA performance in a zero-shot setting.

Findings

LAMOC outperforms several zero-shot methods on A-OKVQA dataset.

LAMOC achieves results comparable to fine-tuned VLP models.

The approach effectively integrates captioning and language model feedback for VQA.

Abstract

In this paper, we propose a novel language model guided captioning approach, LAMOC, for knowledge-based visual question answering (VQA). Our approach employs the generated captions by a captioning model as the context of an answer prediction model, which is a Pre-trained Language model (PLM). As the major contribution, we leverage the guidance and feedback of the prediction model to improve the capability of the captioning model. In this way, the captioning model can become aware of the task goal and information need from the PLM. To develop our approach, we design two specific training stages, where the first stage adapts the captioning model to the prediction model (selecting more suitable caption propositions for training) and the second stage tunes the captioning model according to the task goal (learning from feedback of the PLM). Extensive experiments demonstrate the effectiveness of the proposed approach on the knowledge-based VQA task. Specifically, on the challenging A-OKVQA dataset, LAMOC outperforms several competitive zero-shot methods and even achieves comparable results to a fine-tuned VLP model. Our code is publicly available at https://github.com/RUCAIBox/LAMOC.