Generalization Matters: Loss Minima Flattening via Parameter Hybridization for Efficient Online Knowledge Distillation.
Tianli ZhangMengqi XueJiangtao ZhangHaofei ZhangYu WangLechao ChengJie SongMingli SongPublished in: CoRR (2023)
Keyphrases
- online learning
- domain knowledge
- knowledge acquisition
- knowledge extraction
- data mining
- knowledge base
- expert systems
- knowledge representation
- neural network
- computationally efficient
- knowledge management
- prior knowledge
- support vector
- real time
- data mining techniques
- decision trees
- knowledge based systems
- higher level
- social networks
- design process
- learning systems
- background knowledge
- knowledge sources
- parameter values
- input parameters
- efficient learning