CLIP
收藏资源简介:
CLIP数据集是由麻省理工学院创建,旨在从医院出院记录中提取临床行动项,以帮助医生更有效地共享信息。该数据集包含718份由医生标注的文档,覆盖10万句,用于多方面的提取式总结任务,每个方面代表一种需要采取的行动类型。CLIP数据集的应用领域主要集中在提高患者安全和医生效率,通过自动提取行动项来减少医生在电子健康记录系统中的工作负担。
The CLIP dataset was developed by the Massachusetts Institute of Technology (MIT) to extract clinical action items from hospital discharge records, thereby enabling more efficient information sharing among physicians. This dataset includes 718 physician-annotated documents totaling over 100,000 sentences, and is designed for multi-facet extractive summarization tasks, where each facet corresponds to a distinct type of actionable clinical requirement. The primary applications of the CLIP dataset focus on improving patient safety and physician efficiency, by reducing the administrative workload of physicians working with electronic health record (EHR) systems through automated extraction of clinical action items.

- CLIP数据集首次发表于OpenAI的研究论文《Learning Transferable Visual Models From Natural Language Supervision》,标志着多模态学习领域的重要突破。
- CLIP数据集开始被广泛应用于图像分类、图像生成和自然语言处理等多个领域,展示了其在跨模态任务中的强大潜力。
- 随着研究的不断深入,CLIP数据集在多个国际会议和期刊上被引用和讨论,进一步推动了多模态学习技术的发展。
- 1Learning Transferable Visual Models From Natural Language SupervisionOpenAI · 2021年
- 2Zero-Shot Text-to-Image GenerationOpenAI · 2021年
- 3CLIP: Connecting Text and ImagesOpenAI · 2021年
- 4Multimodal Neurons in Artificial Neural NetworksOpenAI · 2021年
- 5CLIP-Guided Diffusion Models for Robust Image ManipulationUniversity of California, Berkeley · 2022年



