Novel Meta-Learning Techniques for the Multiclass Image Classification Problem
Multiclass image classification is a complex task that has been thoroughly investigated in the past. Decomposition-based strategies are commonly employed to address it. Typically, these methods divide the original problem into smaller, potentially simpler problems, allowing the application of numero...
| Main Authors: | , , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
MDPI AG
2022-12-01
|
| Series: | Sensors |
| Subjects: | |
| Online Access: | https://www.mdpi.com/1424-8220/23/1/9 |
| _version_ | 1797431196222750720 |
|---|---|
| author | Antonios Vogiatzis Stavros Orfanoudakis Georgios Chalkiadakis Konstantia Moirogiorgou Michalis Zervakis |
| author_facet | Antonios Vogiatzis Stavros Orfanoudakis Georgios Chalkiadakis Konstantia Moirogiorgou Michalis Zervakis |
| author_sort | Antonios Vogiatzis |
| collection | DOAJ |
| description | Multiclass image classification is a complex task that has been thoroughly investigated in the past. Decomposition-based strategies are commonly employed to address it. Typically, these methods divide the original problem into smaller, potentially simpler problems, allowing the application of numerous well-established learning algorithms that may not apply directly to the original task. This work focuses on the efficiency of decomposition-based methods and proposes several improvements to the meta-learning level. In this paper, four methods for optimizing the ensemble phase of multiclass classification are introduced. The first demonstrates that employing a mixture of experts scheme can drastically reduce the number of operations in the training phase by eliminating redundant learning processes in decomposition-based techniques for multiclass problems. The second technique for combining learner-based outcomes relies on Bayes’ theorem. Combining the Bayes rule with arbitrary decompositions reduces training complexity relative to the number of classifiers even further. Two additional methods are also proposed for increasing the final classification accuracy by decomposing the initial task into smaller ones and ensembling the output of the base learners along with that of a multiclass classifier. Finally, the proposed novel meta-learning techniques are evaluated on four distinct datasets of varying classification difficulty. In every case, the proposed methods present a substantial accuracy improvement over existing traditional image classification techniques. |
| first_indexed | 2024-03-09T09:41:25Z |
| format | Article |
| id | doaj.art-2185710ce1a34d03b21a5c6e0fd31e24 |
| institution | Directory Open Access Journal |
| issn | 1424-8220 |
| language | English |
| last_indexed | 2024-03-09T09:41:25Z |
| publishDate | 2022-12-01 |
| publisher | MDPI AG |
| record_format | Article |
| series | Sensors |
| spelling | doaj.art-2185710ce1a34d03b21a5c6e0fd31e242023-12-02T00:52:30ZengMDPI AGSensors1424-82202022-12-01231910.3390/s23010009Novel Meta-Learning Techniques for the Multiclass Image Classification ProblemAntonios Vogiatzis0Stavros Orfanoudakis1Georgios Chalkiadakis2Konstantia Moirogiorgou3Michalis Zervakis4School of ECE, Technical University of Crete, 731 00 Chania, GreeceSchool of ECE, Technical University of Crete, 731 00 Chania, GreeceSchool of ECE, Technical University of Crete, 731 00 Chania, GreeceSchool of ECE, Technical University of Crete, 731 00 Chania, GreeceSchool of ECE, Technical University of Crete, 731 00 Chania, GreeceMulticlass image classification is a complex task that has been thoroughly investigated in the past. Decomposition-based strategies are commonly employed to address it. Typically, these methods divide the original problem into smaller, potentially simpler problems, allowing the application of numerous well-established learning algorithms that may not apply directly to the original task. This work focuses on the efficiency of decomposition-based methods and proposes several improvements to the meta-learning level. In this paper, four methods for optimizing the ensemble phase of multiclass classification are introduced. The first demonstrates that employing a mixture of experts scheme can drastically reduce the number of operations in the training phase by eliminating redundant learning processes in decomposition-based techniques for multiclass problems. The second technique for combining learner-based outcomes relies on Bayes’ theorem. Combining the Bayes rule with arbitrary decompositions reduces training complexity relative to the number of classifiers even further. Two additional methods are also proposed for increasing the final classification accuracy by decomposing the initial task into smaller ones and ensembling the output of the base learners along with that of a multiclass classifier. Finally, the proposed novel meta-learning techniques are evaluated on four distinct datasets of varying classification difficulty. In every case, the proposed methods present a substantial accuracy improvement over existing traditional image classification techniques.https://www.mdpi.com/1424-8220/23/1/9ensemble learningmixture of expertsdecomposition-based methodsmulti-class classificationBayes ruleopinion aggregation |
| spellingShingle | Antonios Vogiatzis Stavros Orfanoudakis Georgios Chalkiadakis Konstantia Moirogiorgou Michalis Zervakis Novel Meta-Learning Techniques for the Multiclass Image Classification Problem Sensors ensemble learning mixture of experts decomposition-based methods multi-class classification Bayes rule opinion aggregation |
| title | Novel Meta-Learning Techniques for the Multiclass Image Classification Problem |
| title_full | Novel Meta-Learning Techniques for the Multiclass Image Classification Problem |
| title_fullStr | Novel Meta-Learning Techniques for the Multiclass Image Classification Problem |
| title_full_unstemmed | Novel Meta-Learning Techniques for the Multiclass Image Classification Problem |
| title_short | Novel Meta-Learning Techniques for the Multiclass Image Classification Problem |
| title_sort | novel meta learning techniques for the multiclass image classification problem |
| topic | ensemble learning mixture of experts decomposition-based methods multi-class classification Bayes rule opinion aggregation |
| url | https://www.mdpi.com/1424-8220/23/1/9 |
| work_keys_str_mv | AT antoniosvogiatzis novelmetalearningtechniquesforthemulticlassimageclassificationproblem AT stavrosorfanoudakis novelmetalearningtechniquesforthemulticlassimageclassificationproblem AT georgioschalkiadakis novelmetalearningtechniquesforthemulticlassimageclassificationproblem AT konstantiamoirogiorgou novelmetalearningtechniquesforthemulticlassimageclassificationproblem AT michaliszervakis novelmetalearningtechniquesforthemulticlassimageclassificationproblem |