Saturated Post-hoc Optimization for Classical Planning

被引:0
作者
Seipp, Jendrik [1 ,2 ]
Keller, Thomas [1 ]
Helmert, Malte [1 ]
机构
[1] Univ Basel, Basel, Switzerland
[2] Linkoping Univ, Linkoping, Sweden
来源
THIRTY-FIFTH AAAI CONFERENCE ON ARTIFICIAL INTELLIGENCE, THIRTY-THIRD CONFERENCE ON INNOVATIVE APPLICATIONS OF ARTIFICIAL INTELLIGENCE AND THE ELEVENTH SYMPOSIUM ON EDUCATIONAL ADVANCES IN ARTIFICIAL INTELLIGENCE | 2021年 / 35卷
基金
欧盟地平线“2020”; 欧洲研究理事会;
关键词
D O I
暂无
中图分类号
TP18 [人工智能理论];
学科分类号
081104 ; 0812 ; 0835 ; 1405 ;
摘要
Saturated cost partitioning and post-hoc optimization are two powerful cost partitioning algorithms for optimal classical planning. The main idea of saturated cost partitioning is to give each considered heuristic only the fraction of remaining operator costs that it needs to prove its estimates. We show how to apply this idea to post-hoc optimization and obtain a heuristic that dominates the original both in theory and on the IPC benchmarks.
引用
收藏
页码:11947 / 11953
页数:7
相关论文
共 23 条
[1]  
[Anonymous], 2008, P 18 INT C AUT PLANN
[2]  
[Anonymous], 1998, Theory of Linear and Integer Programming
[3]  
Edelkamp S., 2001, EUROPEAN C PLANNING
[4]  
Edelkamp S., 2006, P 4 WORKSH MOD CHECK, P35
[5]   Additive pattern database heuristics [J].
Felner, A ;
Korf, RE ;
Hanan, S .
JOURNAL OF ARTIFICIAL INTELLIGENCE RESEARCH, 2004, 22 :279-318
[6]  
Haslum P, 2005, P 20 NAT C ART INT A, V20, P1163
[7]  
Haslum P., 2007, Domain-Independent Construction of Pattern Database Heuristics for Cost-Optimal Planning, P1007
[8]  
Helmert M., 2007, ICAPS, P176
[9]   The Fast Downward planning system [J].
Helmert, Malte .
JOURNAL OF ARTIFICIAL INTELLIGENCE RESEARCH, 2006, 26 :191-246
[10]   Merge-and-Shrink Abstraction: A Method for Generating Lower Bounds in Factored State Spaces [J].
Helmert, Malte ;
Haslum, Patrik ;
Hoffmann, Joerg ;
Nissim, Raz .
JOURNAL OF THE ACM, 2014, 61 (03)