Timezone: »

RLEG: Vision-Language Representation Learning with Diffusion-based Embedding Generation
Liming Zhao · Kecheng Zheng · Yun Zheng · Deli Zhao · Jingren Zhou

Wed Jul 26 05:00 PM -- 06:30 PM (PDT) @ Exhibit Hall 1 #805

Vision-language representation learning models (e.g., CLIP) have achieved state-of-the-art performance on various downstream tasks, which usually need large-scale training data to learn discriminative representation. Recent progress on generative diffusion models (e.g., DALL-E 2) has demonstrated that diverse high-quality samples can be synthesized by randomly sampling from generative distribution. By virtue of generative capability in this paper, we propose a novel vision-language Representation Learning method with diffusion-based Embedding Generation (RLEG), which exploits diffusion models to generate feature embedding online for learning effective vision-language representation. Specifically, we first adopt image and text encoders to extract the corresponding embeddings. Secondly, pretrained diffusion-based embedding generators are harnessed to transfer the embedding modality online between vision and language domains. The embeddings generated from the generators are then served as augmented embedding-level samples, which are applied to contrastive learning with the variant of the CLIP framework. Experimental results show that the proposed method could learn effective representation and achieve state-of-the-art performance on various tasks including image classification, image-text retrieval, object detection, semantic segmentation, and text-conditional image generation.

Author Information

Liming Zhao (Alibaba Group)
Liming Zhao

Liming Zhao is currently a research scientist at Alibaba DAMO Academy. His current research interests are primarily in computer vision and machine learning, especially large-scale multimodal foundation model pretraining. His research works have been published in journals/conferences such as TPAMI, TIP, CVPR, ICCV, ECCV, ICML, AAAI and IJCAI. He serves as a reviewer for top journals (IEEE TIP and ACM TIST) and conferences (CVPR, ICCV, AAAI, IJCAI) .

Kecheng Zheng (Ant Research)
Yun Zheng (Alibaba Group)
Deli Zhao (Alibaba Group)
Jingren Zhou (Alibaba Group)

More from the Same Authors