Knowledge Commons of Institute of Automation,CAS
How to Make Cross Encoder a Good Teacher for Efficient Image-Text Retrieval? | |
chen yuxin1,2,3![]() ![]() ![]() ![]() ![]() | |
2024-06 | |
会议名称 | The IEEE/CVF Conference of Computer Vision and Pattern Recognition |
会议日期 | 2024-6 |
会议地点 | 美国西雅图 |
摘要 | Dominant dual-encoder models enable efficient imagetext retrieval but suffer from limited accuracy, while the cross-encoder models offer higher accuracy at the expense of efficiency. Distilling cross-modality matching knowledge from cross-encoder to dual-encoder provides a natural approach to harness their strengths. Thus, we investigate the following valuable question: how to make crossencoder a good teacher for dual-encoder? Our findings are threefold: (1) Cross-modal similarity score distribution of cross-encoder is more concentrated, while the result of dual-encoder is nearly normal, making vanilla logit distillation less effective. However, ranking distillation remains practical, as it is not affected by the score distribution. (2) Only the relative order between hard negatives conveys valid knowledge, while the order information between easy negatives has little significance. (3) Maintaining the coordination between distillation loss and dual-encoder training loss is beneficial for knowledge transfer. Based on these findings, we propose a novel Contrastive Partial Ranking Distillation (CPRD) method, which implements the objective of mimicking relative order between hard negative samples with contrastive learning. This approach coordinates with the training of the dual-encoder, effectively transferring valid knowledge from the cross-encoder to the dualencoder. Extensive experiments on image-text retrieval and ranking tasks show that our method surpasses other distillation methods and significantly improves the accuracy of dual-encoder. |
收录类别 | EI |
七大方向——子方向分类 | 多模态智能 |
国重实验室规划方向分类 | 多模态协同认知 |
是否有论文关联数据集需要存交 | 否 |
文献类型 | 会议论文 |
条目标识符 | http://ir.ia.ac.cn/handle/173211/57582 |
专题 | 多模态人工智能系统全国重点实验室_视频内容安全 |
通讯作者 | yuan chunfeng |
作者单位 | 1.State Key Laboratory of Multimodal Artificial Intelligence Systems, Institute of Automation, Chinese Academy of Sciences 2.ARC Lab, Tencent PCG 3.School of Artificial Intelligence, University of Chinese Academy of Sciences 4.School of Information Science and Technology, ShanghaiTech University 5.The University of Hong Kong |
第一作者单位 | 中国科学院自动化研究所 |
通讯作者单位 | 中国科学院自动化研究所 |
推荐引用方式 GB/T 7714 | chen yuxin,ma zongyang,zhang ziqi,et al. How to Make Cross Encoder a Good Teacher for Efficient Image-Text Retrieval?[C],2024. |
条目包含的文件 | 下载所有文件 | |||||
文件名称/大小 | 文献类型 | 版本类型 | 开放类型 | 使用许可 | ||
Chen_How_to_Make_Cro(1070KB) | 会议论文 | 开放获取 | CC BY-NC-SA | 浏览 下载 |
除非特别说明,本系统中所有内容都受版权保护,并保留所有权利。
修改评论