重新审视视觉-语言模型中语言先验的作用
Revisiting the Role of Language Priors in Vision-Language Models
AI总结：
该研究探究生成式视觉-语言模型的零样本图文检索性能，提出VisualGPTScore匹配分数，发现语言偏致部分基准性能偏差，给出无需微调的概率去偏后处理方案，提升了零样本基线性能。
摘要：
视觉-语言模型（VLM）具有广泛影响力，部分原因在于它们无需任何微调，就能以零样本方式应用于各类视觉理解任务。我们研究以图像为输入、训练用于下一词生成的生成式VLM（generative VLM），并在8个主流视觉-语言基准上，以图文检索这一典型任务探究其零样本性能。我们的首个发现是，只需计算给定图像时生成特定文本串的匹配分数，就能将生成式VLM改造用于判别式任务（如图文检索），我们将这一概率分数命名为视觉生成预训练分数（VisualGPTScore）。VisualGPTScore在部分检索基准上取得接近完美的准确率，但在另一些基准上准确率不佳。我们从概率角度分析这一现象，指出部分基准通过构造具有对抗性但不符合真实分布的文本描述，无意中捕获了非自然的语言分布。事实上，我们证明，即使是忽略所有图像信息的“盲”语言模型，有时也能超越此前所有最优方法，这让人联想到多年前视觉问答（VQA）领域曾面临的类似挑战。我们提出一种概率后处理方案，无需重新训练或微调模型，就能在测试时控制生成式VLM中的语言偏置程度。实验表明，经过适当去偏的VisualGPTScore是视觉-语言理解领域的强零样本基线，通常能达到当前最优准确率。
英文摘要：
Vision-language models (VLMs) are impactful in part because they can be applied to a variety of visual understanding tasks in a zero-shot fashion, without any fine-tuning. We study $\textit{generative VLMs}$ that are trained for next-word generation given an image. We explore their zero-shot performance on the illustrative task of image-text retrieval across 8 popular vision-language benchmarks. Our first observation is that they can be repurposed for discriminative tasks (such as image-text retrieval) by simply computing the match score of generating a particular text string given an image. We call this probabilistic score the $\textit{Visual Generative Pre-Training Score}$ (VisualGPTScore). While the VisualGPTScore produces near-perfect accuracy on some retrieval benchmarks, it yields poor accuracy on others. We analyze this behavior through a probabilistic lens, pointing out that some benchmarks inadvertently capture unnatural language distributions by creating adversarial but unlikely text captions. In fact, we demonstrate that even a "blind" language model that ignores any image evidence can sometimes outperform all prior art, reminiscent of similar challenges faced by the visual-question answering (VQA) community many years ago. We derive a probabilistic post-processing scheme that controls for the amount of linguistic bias in generative VLMs at test time without having to retrain or fine-tune the model. We show that the VisualGPTScore, when appropriately debiased, is a strong zero-shot baseline for vision-language understanding, oftentimes producing state-of-the-art accuracy.
机构、总结、中文摘要均由AI分析而来，可能存在错误，请以论文正文为准。
