MLSP-L22: Deep Learning Training Methods I
Fri, 19 Apr, 13:10 - 15:10 (UTC +9)
Location: Room 103
Session Type: Lecture
Session Co-Chairs: Kai Yu, Shanghai Jiao Tong University and Qiuqiang Kong, Chinese University of Hong Kong
Track: Machine Learning for Signal Processing
Click the to view the manuscript on IEEE Xplore Open Preview
Fri, 19 Apr, 13:10 - 13:30 (UTC +9)
 

MLSP-L22.1: LACVIT: A LABEL-AWARE CONTRASTIVE FINE-TUNING FRAMEWORK FOR VISION TRANSFORMERS
Zijun Long, Richard McCreadie, Gerardo Aragon Camarasa, Zaiqiao Meng, The University of Glasgow, United Kingdom of Great Britain and Northern Ireland
Fri, 19 Apr, 13:30 - 13:50 (UTC +9)
 

MLSP-L22.2: STREAMING ANCHOR LOSS: AUGMENTING SUPERVISION WITH TEMPORAL SIGNIFICANCE
Utkarsh (Oggy) Sarawgi, John Berkowitz, Vineet Garg, Arnav Kundu, Minsik Cho, Sai Srujana Buddi, Saurabh Adya, Ahmed Tewfik, Apple, United States of America
Fri, 19 Apr, 13:50 - 14:10 (UTC +9)
 

MLSP-L22.3: Continuous Review and Timely Correction: Enhancing the Resistance to Noisy Labels via Self-Not-True Distillation
Jingyi Wang, Da Huang, National University of Defense Technology, China; Xinghao Wu, Beihang University, China; Yuhua Tang, Long Lan, National University of Defense Technology, China
Fri, 19 Apr, 14:10 - 14:30 (UTC +9)
 

MLSP-L22.4: SEMANTIC-ENHANCED SUPERVISED CONTRASTIVE LEARNING
Pingyue Zhang, Mengyue Wu, Kai Yu, Shanghai Jiao Tong University, China
Fri, 19 Apr, 14:30 - 14:50 (UTC +9)
 

MLSP-L22.5: AdaPlus: Integrating Nesterov Momentum and Precise Stepsize Adjustment on AdamW Basis
Lei Guan, National University of Defense Technology, China
Fri, 19 Apr, 14:50 - 15:10 (UTC +9)
 

MLSP-L22.6: MULTI-TEACHER DISTILLATION FOR INCREMENTAL OBJECT DETECTION
Le Jiang, Hongqiang Cheng, Xiaozhou Ye, Ye Ouyang, AsiaInfo Technologies (China), China