Changes

Jump to: navigation, search

Uncertainty Cut-Offs

311 bytes removed, 11:59, 17 January 2019
no edit summary
'''Uncertainty Cut-Offs''',<br/>
a safe pruning technique performed in [[Principal Variation Search|PVS]] at [[Sibling Node|siblings]] of a [[Node Types#PV|PV-node]], which are [[Node Types#CUT|Cut-nodes]], expected to [[Fail-High|fail-high]] and therefor to don't improve the score at the parent PV-node. Even more, the fail-high occurs in > 90% from the first move <ref>[[Aske Plaat]], [[Jonathan Schaeffer]], [[Wim Pijls]], [[Arie de Bruin]] ('''1996'''). ''Exploiting Graph Properties of Game Trees.'' [http://www.aaai.org/[Conferences/#AAAI/aaai96.php 13th National Conference on Artificial Intelligence] ([http://www.aaai.org/Press/Proceedings/aaai96.php -96|AAAI-96]), Vol. 1, pp. 234-239, ISBN 978-0-262-51091-2. [http://citeseerx.ist.psu.edu/viewdoc/download;jsessionid=812B9CDE7707A402265713DD254EE227?doi=10.1.1.50.3346&rep=rep1&type=pdf pdf]</ref> . Sometimes, previous expectations are changing, and an expected Cut-node turns out to become an [[Node Types#ALL|All-node]], yielding in a re-search.
=The Idea=

Navigation menu