Improving the learning efficiencies of realtime search
- Kyoto Univ. (Japan)
The capability of learning is one of the salient features of realtime search algorithms such as LRTA*. The major impediment is, however, the instability of the solution quality during convergence: (1) they try to find all optimal solutions even after obtaining fairly good solutions, and (2) they tend to move towards unexplored areas thus failing to balance exploration and exploitation. We propose and analyze two new realtime search algorithms to stabilize the convergence process. {epsilon}-search (weighted realtime search) allows suboptimal solutions with {epsilon} error to reduce the total amount of learning performed. {delta}-search (realtime search with upper bounds) utilizes the upper bounds of estimated costs, which become available after the problem is solved once. Guided by the upper bounds, {delta}-search can better control the tradeoff between exploration and exploitation.
- OSTI ID:
- 430672
- Report Number(s):
- CONF-960876-; TRN: 96:006521-0047
- Resource Relation:
- Conference: 13. National conference on artifical intelligence and the 8. Innovative applications of artificial intelligence conference, Portland, OR (United States), 4-8 Aug 1996; Other Information: PBD: 1996; Related Information: Is Part Of Proceedings of the thirteenth national conference on artificial intelligence and the eighth innovative applications of artificial intelligence conference. Volume 1 and 2; PB: 1626 p.
- Country of Publication:
- United States
- Language:
- English
Similar Records
A Novel Active Optimization Approach for Rapid and Efficient Design Space Exploration Using Ensemble Machine Learning
When Do Extended Physics-Informed Neural Networks (XPINNs) Improve Generalization?