arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2102.10407cs.CVcs.AIcs.CLcs.MM
收藏

VisualGPT:用于图像描述的预训练语言模型的数据高效自适应

VisualGPT: Data-efficient Adaptation of Pretrained Language Models for Image Captioning

King Abdullah University of Science and Technology (KAUST)(阿卜杜拉国王科技大学);Carnegie Mellon University(卡内基梅隆大学);Nanyang Technological University(南洋理工大学)

Jun Chen, Han Guo, Kai Yi, Boyang Li, Mohamed Elhoseiny

摘要:

从少量训练数据中快速学习的能力拓宽了机器学习的应用范围。在本文中,我们提出了一个数据高效的图像描述模型VisualGPT,它利用了来自大型预训练语言模型(LM)的语言知识。一个关键的挑战是在图像中视觉信息的使用和从预训练中获得的先前语言知识之间取得平衡。我们设计了一种新的自恢复编码器-解码器注意力机制,在少量域内训练数据上快速地将预训练的LM作为语言解码器。所提出的自恢复激活单元产生稀疏激活,但具有降低的对零梯度的敏感性。我们训练提出的模型,VisualGPT,0.1%,0.5%和1%的MSCOCO和概念说明训练数据。在这些条件下,我们在MS COCO上的表现优于最佳基线模型高达10.8%CIDEr,在概念字幕上的表现优于最佳基线模型高达5.4%CIDEr。此外,Visual-GPT在IU X射线上实现了最先进的结果,这是一个医疗报告生成数据集。据我们所知,这是第一个工作,提高数据效率的图像描述通过利用LM预训练的单峰数据。我们的代码可在https://github.com/Vision-CAIR/VisualGPT上获取。

英文摘要:

The ability to quickly learn from a small quantity oftraining data widens the range of machine learning applications. In this paper, we propose a data-efficient image captioning model, VisualGPT, which leverages the linguistic knowledge from a large pretrained language model(LM). A crucial challenge is to balance between the use of visual information in the image and prior linguistic knowledge acquired from pretraining. We designed a novel self-resurrecting encoder-decoder attention mechanism to quickly adapt the pretrained LM as the language decoder ona small amount of in-domain training data. The proposed self-resurrecting activation unit produces sparse activations but has reduced susceptibility to zero gradients. We train the proposed model, VisualGPT, on 0.1%, 0.5% and 1% of MSCOCO and Conceptual Captions training data. Under these conditions, we outperform the best baseline model by up to 10.8% CIDEr on MS COCO and upto 5.4% CIDEr on Conceptual Captions. Further, Visual-GPT achieves the state-of-the-art result on IU X-ray, a medical report generation dataset. To the best of our knowledge, this is the first work that improves data efficiency of image captioning by utilizing LM pretrained on unimodal data. Our code is available at: https://github.com/Vision-CAIR/VisualGPT.

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。

↑