Generalization Matters: Loss Minima Flattening via Parameter Hybridization for Efficient Online Knowledge Distillation.
Tianli ZhangMengqi XueJiangtao ZhangHaofei ZhangYu WangLechao ChengJie SongMingli SongPublished in: CVPR (2023)
Keyphrases
- expert systems
- domain knowledge
- online learning
- knowledge base
- prior knowledge
- computationally expensive
- knowledge representation
- parameter space
- learning systems
- computationally efficient
- machine learning
- social media
- design process
- recommender systems
- domain experts
- background knowledge
- knowledge sharing
- data mining