China
Africa
Explore chapters and articles related to this topic
Parcar: A Parallel Cost-Based Abductive Reasoning System
Published in Takushi Tanaka, Setsuo Ohsuga, Moonis Ali, Industrial and Engineering Applications of Artificial Intelligence and Expert Systems, 2022
Shohei Kato, Chiemi Kamakura, Hirohisa Seki, Hidenori Itoh
This paper proposed a parallelization method of cost-based Horn abduction. We introduced a search control technique of parallel best-first heuristic search into abductive reasoning mechanism, thereby obtaining much more efficiently the optimal explanation of the given observation. We then implemented a parallel cost-based abductive reasoning system PARCAR on an MIMD distributed memory parallel computer Fujitsu AP1000. The well performance results were obtained by some experiments on AP1000. In the experiments, we supposed that h(g) = 0 for all goals g (see definition 2.3), because the experiments were to evaluate the parallelism of the algorithms. We have already proposed the pre-analysis to discover effective admissible heuristic function ĥ(g) in [KSI94]. It can also make PARCAR more efficient, by pruning the search space.
Multi-Aerial-Robot Planning
Published in Yasmina Bestaoui Sebbane, Multi-UAV Planning and Task Allocation, 2020
The algorithm must have an admissible heuristic function that is efficiently computable and as close as possible to the true value (to maximize the amount of pruning). HT-t(s0,δt)=∑s∈SProb(s|s0,δt)hT-t(s)
Petri Net Modeling and Scheduling of Automated Manufacturing Systems
Published in Cornelius Leondes, Optimization Methods for Manufacturing, 2019
MengChu Zhou, Huanxin Henry Xiong
Even though the employed heuristic function is admissible, a more effective admissible heuristic function is desired to reduce the search effort. For hybrid search schemes, instead of employing BT on the top and BF on the bottom or vice versa, a more effective way should be employing BT and BF interchangeably based on the current state. This requires a comprehensive analysis of proposed schemes.
Improving Domain-Independent Heuristic State-Space Planning via plan cost predictions
Published in Journal of Experimental & Theoretical Artificial Intelligence, 2023
Francesco Percassi, Alfonso E. Gerevini, Enrico Scala, Ivan Serina, Mauro Vallati
As our anchor heuristic, we use the admissible heuristic . As a guiding heuristic, we use one of the following inadmissible heuristics: (Hoffmann & Nebel, 2001; Richter & Westphal, 2010). In our experiments we compare the effect of using any combination of the aforementioned inadmissible heuristics by comparing the performance obtained with the same search without the adaption enabled. Figure 2 details the exact configuration of the FD system with the different inadmissible heuristics. For the configuration with more inadmissible heuristics, we run experiments using the adaptive heuristic to either one of them, or both of them. A first set of experiments is devoted to investigate the best set of heuristics to use, and the best heuristic to be modified using the provided prediction . Then we focus our analysis on the best heuristic and report on a domain by domain evaluation.
Decision-dependent distributionally robust Markov decision process method in dynamic epidemic control
Published in IISE Transactions, 2023
Jun Song, William Yang, Chaoyue Zhao
It is worth noting that if is an admissible heuristic function, meaning that its values are less than or equal to the true reward values, the algorithm will converge to the optimal solution. This result is formally stated below.