C4.5: programs for machine learning
C4.5: programs for machine learning
A tight analysis of the greedy algorithm for set cover
STOC '96 Proceedings of the twenty-eighth annual ACM symposium on Theory of computing
Stochastic dynamic programming with factored representations
Artificial Intelligence
Functional Models for Regression Tree Leaves
ICML '97 Proceedings of the Fourteenth International Conference on Machine Learning
Approximation algorithms for combinatorial problems
STOC '73 Proceedings of the fifth annual ACM symposium on Theory of computing
Machine Learning
Causal Graph Based Decomposition of Factored MDPs
The Journal of Machine Learning Research
Hierarchical reinforcement learning with the MAXQ value function decomposition
Journal of Artificial Intelligence Research
Exploiting structure in policy construction
IJCAI'95 Proceedings of the 14th international joint conference on Artificial intelligence - Volume 2
AAAI'96 Proceedings of the thirteenth national conference on Artificial intelligence - Volume 1
A Bayesian approach to learning Bayesian networks with local structure
UAI'97 Proceedings of the Thirteenth conference on Uncertainty in artificial intelligence
Hi-index | 0.05 |
This paper addresses the problem of learning dynamic Bayesian network (DBN) models to support reinforcement learning. It focuses on learning regression tree (context-specific dependence) models of the conditional probability distributions of the DBNs. Existing algorithms rely on standard regression tree learning methods (both propositional and relational). However, such methods presume that the stochasticity in the domain can be modeled as a deterministic function with additive noise. This is inappropriate for many RL domains, where the stochasticity takes the form of stochastic choice over deterministic functions. This paper introduces a regression tree algorithm in which each leaf node is modeled as a finite mixture of deterministic functions. This mixture is approximated via a greedy set cover. Experiments on three challenging RL domains show that this approach finds trees that are more accurate and that are more likely to correctly identify the conditional dependencies in the DBNs based on small samples.