The complexity of Markov decision processes
Mathematics of Operations Research
Artificial Intelligence
Proceedings of the seventh international conference (1990) on Machine learning
Incremental path planning on graphs with cycles
Proceedings of the first international conference on Artificial intelligence planning systems
Efficient learning and planning within the Dyna framework
Proceedings of the second international conference on From animals to animats 2 : simulation of adaptive behavior: simulation of adaptive behavior
Learning continuous-space navigation heuristics in real time
Proceedings of the second international conference on From animals to animats 2 : simulation of adaptive behavior: simulation of adaptive behavior
Memory-Based Reinforcement Learning: Efficient Computation with Prioritized Sweeping
Advances in Neural Information Processing Systems 5, [NIPS Conference]
Complexity Analysis of Real-Time Reinforcement Learning Applied to Finding Shortest Paths in Deterministic Domains
Optimal Probabilistic and Decision-Theoretic Planning using Markovian
Optimal Probabilistic and Decision-Theoretic Planning using Markovian
Learning and Sequential Decision Making
Learning and Sequential Decision Making
Learning in embedded systems
Random walks, universal traversal sequences, and the complexity of maze problems
SFCS '79 Proceedings of the 20th Annual Symposium on Foundations of Computer Science
SFCS '90 Proceedings of the 31st Annual Symposium on Foundations of Computer Science
The Journal of Machine Learning Research
Automatic skill acquisition in reinforcement learning using graph centrality measures
Intelligent Data Analysis
Hi-index | 0.01 |
This paper analyzes the complexity of on-line reinforcement learning algorithms, namely asynchronous realtime versions of Q-learning and value-iteration, applied to the problem of reaching a goal state in deterministic domains. Previous work had concluded that, in many cases, tabula rasa reinforcement learning was exponential for such problems, or was tractable only if the learning algorithm was augmented. We show that, to the contrary, the algorithms are tractable with only a simple change in the task representation or initialization. We provide tight bounds on the worst-case complexity, and show how the complexity is even smaller if the reinforcement learning algorithms have initial knowledge of the topology of the state space or the domain has certain special properties. We also present a novel bidirectional Q-learning algorithm to find optimal paths from all states to a goal state and show that it is no more complex than the other algorithms.