Gradient Coordination for Quantifying and Maximizing Knowledge Transference in Multi-Task Learning.
Xuanhua YangJianxin ZhaoShaoguo LiuLiang WangBo ZhengPublished in: SIGIR (2023)
Keyphrases
- multi task learning
- multi task
- knowledge base
- learning tasks
- gaussian processes
- prior knowledge
- expert systems
- knowledge representation
- knowledge discovery
- knowledge acquisition
- multiple tasks
- multi label image annotation
- learning problems
- high order
- learning models
- transfer learning
- theoretical analysis
- training data