Learning Temporal Distances: Contrastive Successor Features Can Provide a Metric Structure for Decision-Making.
Vivek MyersChongyi ZhengAnca D. DraganSergey LevineBenjamin EysenbachPublished in: CoRR (2024)
Keyphrases
- decision making
- prior knowledge
- unsupervised learning
- structural information
- learning systems
- learning algorithm
- distance measure
- temporal structure
- supervised learning
- co occurrence
- learning tasks
- feature extraction
- learning models
- image features
- low level
- feature space
- euclidean distance
- learning problems
- information content
- metric learning
- hidden variables