Reinforcement Learning for POMDP: Partitioned Rollout and Policy Iteration with Application to Autonomous Sequential Repair Problems

Mathematics

• Partially Observable Markov Decision Process 100%
• Reinforcement Learning 90%
• Policy Iteration 88%
• Repair 68%
• Policy 59%
• Neural Networks 38%
• Distributed Computation 29%
• Rupture 27%
• Look-ahead 27%
• Robot 24%
• Partial Information 24%
• Function Approximation 23%
• Iteration Scheme 23%
• Infinite Horizon 22%
• Classifier 22%
• Architecture 22%
• Beliefs 21%
• Dynamic Programming 20%
• Cost Function 20%
• State Space 17%
• Partial 13%
• Observation 13%
• Formulation 12%
• Simulation 12%
• Class 5%

Engineering & Materials Science

• Reinforcement learning 60%
• Repair 49%
• Pipelines 46%
• Neural networks 36%
• Dynamic programming 29%
• Cost functions 27%
• Classifiers 21%
• Robots 19%