|
Article information
2000 , Volume 5, ¹ 3, p.110-122
Fedotov A.A.
Compilation of search trees in the absence of the comprehensive information about occurrence rate of the objects to be identified
Search trees are intended for the identification of objects in biology, mineralogy etc. The traditional quality criterion of a search tree is the average time of the object identification. It, in its turn, is determined by the probability distribution over the set of objects which is usually not known with certainty. However, some known data about the objects occurrence rate can be used in order to reduce the time consumption for compiling the search trees. The case is considered when the data on the occurrency rate can be represented as some partial order. In this work a method for constructing a search tree close to optimum for thus specified class under the minimax approach is presented. An example of making use of the described algorithm is given.
[full text] Classificator Msc2000:- *92-08 Computational methods
- 92B10 Taxonomy, statistics
- 94A40 Channel models
Classificator Computer Science:- *E.4 Coding and Information Theory
Keywords: taxonomy, search trees, probability distribution of objects, minimax problem, algorithm
Author(s):
Fedotov Aleksei Anatolievich
Address: 630090, Russia, Moscow
E-mail: alexei.fedotov@gmail.com
Bibliography link:
Fedotov A.A. Compilation of search trees in the absence of the comprehensive information about occurrence rate of the objects to be identified // Computational technologies. 2000. V. 5. ¹ 3. P. 110-122
|
|
|
|
