Реферат: Biclustering is an approach to solving data mining problems, which consists in simultaneously grouping rows and columns of a matrix. In this paper, we solve the problem of finding a bicluster of the maximum size, the elements of which should differ from each other by no more than a given value. To solve it, a new local search algorithm has been developed, representing an iterative greedy search. For its implementation, problem-oriented neighborhoods are constructed, different rules for determining the difference of bicluster elements are used. The constructed algorithm is tested on various types of data, the results are compared with the well-known algorithm of Cheng and Church. In all the examples considered, the sizes of the found biclusters are not less than the biclusters of the Cheng and Church algorithm. At the same time, the difference between the elements of bicluster and their average value in most cases is smaller than for the Cheng and Church biclusters.

Библиографическая ссылка: Levanova T. , Khmara I.
A Local Search Algorithm for the Biclustering Problem
Lecture Notes in Computer Science. 2022. V.13217. P.330–344. DOI: 10.1007/978-3-031-16500-9_27 Scopus РИНЦ OpenAlex

Идентификаторы БД:

Scopus:	2-s2.0-85142684434
РИНЦ:	50334907
OpenAlex:	W4313177465

Цитирование в БД:

БД	Цитирований
OpenAlex	1

Альметрики: