论文部分内容阅读
本文提出一种新的加权算法——ＷＳＡ法，它是在Ａ（或Ａ＊）搜索中引入一种新的加权技术。根据［１］中提出的原理，可以把启发式搜索看成某种随机取样过程，所以通过某种统计推断的方法，可以估计出搜索树中各子树包含目标的可能性．然后把某个权值加到不大可能是解路径上节点的估价函数上，从而使搜索集中在最有希望的路径上。在一致ｍ一枝树上；我们证明了这种加权方法可显著提高效率。
This paper presents a new weighting algorithm - the WSA method, which introduces a new weighting technique into A (or A *) search. According to the principle proposed in [1], the heuristic search can be regarded as some kind of random sampling process. Therefore, by some statistical inference method, it is possible to estimate the possibility that each sub-tree in the search tree contains a target. Then adding a weight to a valuation function that is unlikely to be a node on the solution path concentrates the search on the most promising path. In a consistent tree; we show that this weighting method can significantly improve efficiency.