MMSP-L5.4

Preventing Modality Collapse via Category-Guided Transition Regularization

Yunxuan Xue, Chenwei Cui, Xiaoran Yang, Yuan Gao, Jing Xiao, Gui-Song Xia, Wuhan University, China
Session:
MMSP-L5: Cross-Modal Knowledge Distillation and Model Adaptation Oral
Track:
Multimedia Signal Processing [MM]
Location:
Room 131+132
Presentation Time:
Wed, 6 May, 17:30 - 17:50
Presentation
Discussion
Resources
No resources available.
Session MMSP-L5
MMSP-L5.1: PROMPTSID: A SELF-ITERATIVE DISTILLATION FRAMEWORK FOR UNSUPERVISED ADAPTATION OF VISION-LANGUAGE MODELS
Yikai Lin, Xianwei Zhuang, Junbin Zhang, Peking University, China; Chenxing Li, Zikang Huang, Tencent AI Lab, China; Yuexian Zou, Peking University, China
MMSP-L5.2: DART: a Dual-modality Adaptive Representation with divergence Training framework for ZS-CIR
Shunning Liu, Yifan Wang, Tsinghua University, China; Jianing Lin, Beihang University, China; Yufan Wen, Chun Yuan, Tsinghua University, China
MMSP-L5.3: DEMORERANKER: ENHANCING THE IN-CONTEXT LEARNING CAPABILITY OF MULTI-MODAL LARGE MODELS VIA DEMO RANKING
Yuheng Li, Johns Hopkins University, United States of America; Taoyu Wu, iFLYTEK Co., Ltd., Hong Kong; Jiechao Gao, Stanford University, Hong Kong; Wei Zhu, Wenwen Ouyang, University of Hong Kong, Hong Kong
MMSP-L5.4: Preventing Modality Collapse via Category-Guided Transition Regularization
Yunxuan Xue, Chenwei Cui, Xiaoran Yang, Yuan Gao, Jing Xiao, Gui-Song Xia, Wuhan University, China
MMSP-L5.5: APKD: ALIGNED AND PACED KNOWLEDGE DISTILLATION TOWARDS LIGHTWEIGHT HETEROGENEOUS MULTIMODAL EMOTION RECOGNITION
Yujian Sun, Shandong University of Technology, China; Bingtian Qiao, Yiwen Wang, Fuzhou University, China; Shanliang Yang, Shandong University of Technology, China
MMSP-L5.6: Pairing Denoising Enhanced Hash-aware Distillation for Unsupervised Cross-modal Retrieval
Xinsheng Shu, Mingyong Li, Chongqing Normal University, China
Contacts