An ensemble face recognition mechanism based on three-way decisions

Result code in IS VaVaI

<a href="https://www.isvavai.cz/riv?ss=detail&h=RIV%2F61989100%3A27240%2F23%3A10252933" target="_blank" >RIV/61989100:27240/23:10252933 - isvavai.cz</a>

Result on the web

<a href="https://www.sciencedirect.com/science/article/pii/S1319157823000848?via%3Dihub" target="_blank" >https://www.sciencedirect.com/science/article/pii/S1319157823000848?via%3Dihub</a>

DOI - Digital Object Identifier

<a href="http://dx.doi.org/10.1016/j.jksuci.2023.03.016" target="_blank" >10.1016/j.jksuci.2023.03.016</a>

Result language

angličtina

Original language name

An ensemble face recognition mechanism based on three-way decisions

Original language description

The explainable human-computer interaction (HCI) is about designing approaches capable of using cog-nitive characteristics like humans. One such characteristic is human vision and its accuracy. The accuracy measures the trust in that system. Therefore, improving accuracy in the authorization with identification process is a primary concern for a visual-based explainable human-computer interaction system. In this article, we propose a three-way decision based ensembled face recognition mechanism called E3FRM. The E3FRM uses a three-way approach to determine the match cases and the respective worth of the captured image with the match cases. Features are extracted using PCA/FLD, and the ensembled face recognition algorithms utilize the extracted features to process the image. Ensemble Face recognition approaches find the match cases based on a given threshold. Finally, the three-way decision model evaluates the suitabil-ity of the captured image for acceptance, rejection, or deferred cases with a dual verification mechanism. Experimental results on well-known eighteen datasets suggest improvements in commonly used metrics of F1, Accuracy and Recall by up to 0.8% to 12.8%, 1% to 9.6% and 1.2% to 13.9%, respectively, in compar-ison to the state-of-the-art methods available, including SPCA +, ML-EM, FLDA-SVD, DMMA, Fast-DMMA, LU, LPP, TDL, KCFT, RBF + DT, and NMF. Furthermore, the proposed approach is comparatively analyzed with ensembled face recognition methods that result in an outperformed F1, Accuracy and Recall by up to 1.1% to 10.3%, 0.1% to 7.3% and 0.9% to 10.5%, respectively. These results suggest that the proposed model may improve face recognition accuracy and the resulting trust in the machines.(c) 2023 The Author(s). Published by Elsevier B.V. on behalf of King Saud University. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/).

Czech name

—

Czech description

—

Type

J_imp - Article in a specialist periodical, which is included in the Web of Science database

CEP classification

—

OECD FORD branch

20203 - Telecommunications

Project

—

Continuities

S - Specificky vyzkum na vysokych skolach

Publication year

2023

Confidentiality

S - Úplné a pravdivé údaje o projektu nepodléhají ochraně podle zvláštních právních předpisů

Name of the periodical

Journal of King Saud University - Computer and Information Sciences

ISSN

1319-1578

e-ISSN

2213-1248

Volume of the periodical

35

Issue of the periodical within the volume

4

Country of publishing house

SA - THE KINGDOM OF SAUDI ARABIA

Number of pages

13

Pages from-to

196-208

UT code for WoS article

000995091400001

EID of the result in the Scopus database

—