Generating meta-learning tasks to evolve parametric loss for classification learning.
Zhaoyang HaiXiabi LiuYuchen RenNouman Qadeer SoomroPublished in: CoRR (2021)
Keyphrases
- learning tasks
- supervised learning
- learning problems
- machine learning
- learning algorithm
- machine learning algorithms
- meta learning
- transfer learning
- multi task learning
- learning models
- labeled and unlabeled data
- human learning
- learning situations
- kernel learning
- multitask learning
- multi task
- hypothesis space
- learning experience
- classification accuracy
- learning process
- metric learning
- feature vectors
- meta level
- class labels
- kernel methods
- reinforcement learning
- decision trees
- neural network
- semi supervised learning setting
- multi label
- unlabeled data
- labeled data
- unsupervised learning
- image classification
- knowledge discovery
- prior knowledge
- feature space
- support vector
- feature selection
- data mining