Learning Good State and Action Representations for Markov Decision Process via Tensor Decomposition

Keyphrases

• Markov Decision Process 100%
• Action Representation 100%
• Good State 100%
• Tensor Decomposition 100%
• State Representation 100%
• Good Action 100%
• State Action 66%
• Tensor Structure 66%
• Diffusion Distance 33%
• Accurate Approximation 33%
• Cluster State 33%
• Policy Evaluation 33%
• Block Structure 33%
• Function Approximation 33%
• Importance Sampling 33%
• Transition Kernel 33%
• Unsupervised Learning Techniques 33%
• Kernelization 33%
• Downstream Task 33%
• Continuous States 33%
• Rank-one Approximation 33%
• Enabling Function 33%
• Statistical Error Bound 33%
• Natural Tensors 33%
• Process Abstraction 33%
• Tucker Rank 33%
• Action Abstraction 33%

Computer Science

• And-States 100%
• Markov Decision Process 100%
• Structure Tensor 66%
• Approximation (Algorithm) 33%
• Unsupervised Learning 33%
• Structured Block 33%
• Function Approximation 33%
• Importance Sampling 33%
• Rank Approximation 33%