Markov based genetic algorithm (M-GA): To mine frequent sub components from molecular structures

Title

Markov based genetic algorithm (M-GA): To mine frequent sub components from molecular structures

Creator

Kanmani P.; Prakasi O.S.G.; Jayapandian N.

Description

Processing the molecular compounds to identify the internal chemical structure is a challenging task in bio-chemical research. Popular approaches, mine the frequent subcomponents from the molecules with chemical and biological properties represented in the form of feature vector histogram. Though this helps to identify the absence or presence of mined feature, calculating the frequency of every frequent substructure involves sub graph isomorphism test which is an NP-Complete process. To overcome the above mentioned bottleneck we proposed Markov based Genetic algorithm (M-GA) in which the chemical descriptors were considered from two-dimensional representations of molecules that classify chemical compounds using mining significant substructure and generates the binary vector that generate pure active classes, singleton reactors, descriptor sets. This method scales down the process of mining substructures that are statistically significant from huge chemical databases. The results shows that the performance of proposed algorithm is improved compared to the existing algorithms. 2020, Research Trend. All rights reserved.

Source

International Journal on Emerging Technologies, Vol-11, No. 2, pp. 516-523.

Date

2020-01-01

Publisher

Research Trend

Subject

Fitness function; Histogram; Markov model; Memetic Operator; Molecular substructure

Coverage

Kanmani P., Department of Computer Science and Engineering, Christ (Deemed to be University), Bengaluru, Karnataka, India; Prakasi O.S.G., Department of Computer Science and Engineering, Christ (Deemed to be University), Bengaluru, Karnataka, India; Jayapandian N., Department of Computer Science and Engineering, Christ (Deemed to be University), Bengaluru, Karnataka, India

Rights

Restricted Access

Relation

ISSN: 9758364

Format

Online

Language

English

Type

Article

Identifier

https://www.scopus.com/inward/record.uri?eid=2-s2.0-85084357037&partnerID=40&md5=26fa90ec2cd0a3ba53f060da00bce614