Ensemble of classifiers based on CNN for increasing generalization ability in face image recognition
The paper considers the problem of increasing the generalization ability of classification systems by creating an ensemble of classifiers based on the CNN architecture. Different structures of the ensemble will be considered and compared. Deep learning fulfills an important role in the developed sys...
| Main Authors: | , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
Polish Academy of Sciences
2022-05-01
|
| Series: | Bulletin of the Polish Academy of Sciences: Technical Sciences |
| Subjects: | |
| Online Access: | https://journals.pan.pl/Content/123185/PDF/2636_BPASTS_2022_70_3.pdf |
| _version_ | 1811313240607555584 |
|---|---|
| author | Robert Szmurło Stanislaw Osowski |
| author_facet | Robert Szmurło Stanislaw Osowski |
| author_sort | Robert Szmurło |
| collection | DOAJ |
| description | The paper considers the problem of increasing the generalization ability of classification systems by creating an ensemble of classifiers based on the CNN architecture. Different structures of the ensemble will be considered and compared. Deep learning fulfills an important role in the developed system. The numerical descriptors created in the last locally connected convolution layer of CNN flattened to the form of a vector, are subjected to a few different selection mechanisms. Each of them chooses the independent set of features, selected according to the applied assessment techniques. Their results are combined with three classifiers: softmax, support vector machine, and random forest of the decision tree. All of them do simultaneously the same classification task. Their results are integrated into the final verdict of the ensemble. Different forms of arrangement of the ensemble are considered and tested on the recognition of facial images. Two different databases are used in experiments. One was composed of 68 classes of greyscale images and the second of 276 classes of color images. The results of experiments have shown high improvement of class recognition resulting from the application of the properly designed ensemble. |
| first_indexed | 2024-04-13T10:51:23Z |
| format | Article |
| id | doaj.art-9274f7975b764c7ca9bba1b210d6901e |
| institution | Directory Open Access Journal |
| issn | 2300-1917 |
| language | English |
| last_indexed | 2024-04-13T10:51:23Z |
| publishDate | 2022-05-01 |
| publisher | Polish Academy of Sciences |
| record_format | Article |
| series | Bulletin of the Polish Academy of Sciences: Technical Sciences |
| spelling | doaj.art-9274f7975b764c7ca9bba1b210d6901e2022-12-22T02:49:39ZengPolish Academy of SciencesBulletin of the Polish Academy of Sciences: Technical Sciences2300-19172022-05-01703https://doi.org/10.24425/bpasts.2022.141004Ensemble of classifiers based on CNN for increasing generalization ability in face image recognitionRobert Szmurło0https://orcid.org/0000-0001-8041-4438Stanislaw Osowski1https://orcid.org/0000-0003-3194-4656Faculty of Electrical Engineering, Warsaw University of Technology, Koszykowa 75, 00-662 Warszawa, PolandFaculty of Electronic Engineering, Military University of Technology, gen. S. Kaliskiego 2, 00-908 Warszawa, PolandThe paper considers the problem of increasing the generalization ability of classification systems by creating an ensemble of classifiers based on the CNN architecture. Different structures of the ensemble will be considered and compared. Deep learning fulfills an important role in the developed system. The numerical descriptors created in the last locally connected convolution layer of CNN flattened to the form of a vector, are subjected to a few different selection mechanisms. Each of them chooses the independent set of features, selected according to the applied assessment techniques. Their results are combined with three classifiers: softmax, support vector machine, and random forest of the decision tree. All of them do simultaneously the same classification task. Their results are integrated into the final verdict of the ensemble. Different forms of arrangement of the ensemble are considered and tested on the recognition of facial images. Two different databases are used in experiments. One was composed of 68 classes of greyscale images and the second of 276 classes of color images. The results of experiments have shown high improvement of class recognition resulting from the application of the properly designed ensemble.https://journals.pan.pl/Content/123185/PDF/2636_BPASTS_2022_70_3.pdfcnnensemble of classifiersface recognitionfeature selection |
| spellingShingle | Robert Szmurło Stanislaw Osowski Ensemble of classifiers based on CNN for increasing generalization ability in face image recognition Bulletin of the Polish Academy of Sciences: Technical Sciences cnn ensemble of classifiers face recognition feature selection |
| title | Ensemble of classifiers based on CNN for increasing generalization ability in face image recognition |
| title_full | Ensemble of classifiers based on CNN for increasing generalization ability in face image recognition |
| title_fullStr | Ensemble of classifiers based on CNN for increasing generalization ability in face image recognition |
| title_full_unstemmed | Ensemble of classifiers based on CNN for increasing generalization ability in face image recognition |
| title_short | Ensemble of classifiers based on CNN for increasing generalization ability in face image recognition |
| title_sort | ensemble of classifiers based on cnn for increasing generalization ability in face image recognition |
| topic | cnn ensemble of classifiers face recognition feature selection |
| url | https://journals.pan.pl/Content/123185/PDF/2636_BPASTS_2022_70_3.pdf |
| work_keys_str_mv | AT robertszmurło ensembleofclassifiersbasedoncnnforincreasinggeneralizationabilityinfaceimagerecognition AT stanislawosowski ensembleofclassifiersbasedoncnnforincreasinggeneralizationabilityinfaceimagerecognition |