Learning to act using real-time dynamic programming
Artificial Intelligence
Symbolic generalization for on-line planning
UAI'03 Proceedings of the Nineteenth conference on Uncertainty in Artificial Intelligence
| Hi-index | 0.00 |
ARRAY(0x8475700) to induce abstract classes of state tasks. The second approach seeks to learn such state classes by constructing hierarchical connectionist networks whose units act as abstract features or concepts. Both approaches are designed to facilitate control over memory resources, allowing learning to accelerate from early rote memorization to more globally-scaled generalization.