DSelect-k: Differentiable Selection in the Mixture of Experts with Applications to Multi-Task Learning.
Hussein HazimehZhe ZhaoAakanksha ChowdheryMaheswaran SathiamoorthyYihua ChenRahul MazumderLichan HongEd H. ChiPublished in: NeurIPS (2021)
Keyphrases
- classification models
- multi task learning
- learning models
- dirichlet process
- multi task
- multiple tasks
- learning tasks
- multitask learning
- gaussian processes
- multi label image annotation
- learning problems
- transfer learning
- theoretical analysis
- learning algorithm
- high order
- mixture model
- closed form
- learning experience
- higher order
- multi class