Neural networks with a self-refreshing memory: knowledge transfer in sequential learning tasks without catastrophic forgetting.
Bernard AnsStéphane RoussetPublished in: Connect. Sci. (2000)
Keyphrases
- learning tasks
- knowledge transfer
- neural network
- transfer learning
- machine learning
- learning problems
- multi task
- learning algorithm
- supervised learning
- learning experience
- knowledge construction
- function approximators
- cross domain
- machine learning algorithms
- multi label
- multitask learning
- labeled data
- hypothesis space
- multi task learning
- reinforcement learning
- kernel methods
- knowledge sharing
- training examples
- text categorization
- text classification
- data analysis
- real world