RandMixAugment: A Novel Unified Technique for Region- and Image-Level Data Augmentations
Deep learning models learn powerful representational spaces required for handling complex tasks. Recently, data augmentation techniques, region-level, and image-level augmentation have proved effective in significantly improving deep learning models’ generalization performance. Neverthele...
| Main Authors: | , , , , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
IEEE
2024-01-01
|
| Series: | IEEE Access |
| Subjects: | |
| Online Access: | https://ieeexplore.ieee.org/document/10217802/ |
| _version_ | 1827376418247409664 |
|---|---|
| author | Yosoeb Shin Vikas Palakonda Sangseok Yun Il-Min Kim Seon-Gon Kim Sang-Mi Park Jae-Mo Kang |
| author_facet | Yosoeb Shin Vikas Palakonda Sangseok Yun Il-Min Kim Seon-Gon Kim Sang-Mi Park Jae-Mo Kang |
| author_sort | Yosoeb Shin |
| collection | DOAJ |
| description | Deep learning models learn powerful representational spaces required for handling complex tasks. Recently, data augmentation techniques, region-level, and image-level augmentation have proved effective in significantly improving deep learning models’ generalization performance. Nevertheless, such methods may lose some critical features or are still computationally heavy (or inefficient) due to additional computation burdens. To address this issue, in this paper, we present a novel unified data augmentation method for deep learning models, namely, RandMixAugment, which effectively combines the intrinsic properties of region-level augmentation and image-level augmentation. Specifically, the proposed RandMixAugment employs automated augmentation with masking and mixing operations. Experiments are conducted on well-known CIFAR datasets (CIFAR-10 and CIFAR-100) to verify the effectiveness of the proposed scheme compared to state-of-the-art augmentation techniques. The experimental results demonstrate that the proposed RandMixAugment yields superior performance over state-of-the-art techniques on image classification tasks and further improves the performance of the baseline deep learning model by 1.2% and 2.4% on CIFAR-10 and CIFAR-100 datasets, respectively. |
| first_indexed | 2024-03-08T12:09:36Z |
| format | Article |
| id | doaj.art-58bf71ee5c204a418053071cfbe4ead6 |
| institution | Directory Open Access Journal |
| issn | 2169-3536 |
| language | English |
| last_indexed | 2024-03-08T12:09:36Z |
| publishDate | 2024-01-01 |
| publisher | IEEE |
| record_format | Article |
| series | IEEE Access |
| spelling | doaj.art-58bf71ee5c204a418053071cfbe4ead62024-01-23T00:05:46ZengIEEEIEEE Access2169-35362024-01-01128187819710.1109/ACCESS.2023.330538510217802RandMixAugment: A Novel Unified Technique for Region- and Image-Level Data AugmentationsYosoeb Shin0https://orcid.org/0000-0002-0406-9879Vikas Palakonda1Sangseok Yun2https://orcid.org/0000-0002-7961-1394Il-Min Kim3https://orcid.org/0000-0003-1339-8057Seon-Gon Kim4Sang-Mi Park5Jae-Mo Kang6https://orcid.org/0000-0002-8181-5994Department of Artificial Intelligence, Kyungpook National University, Daegu, South KoreaDepartment of Artificial Intelligence, Kyungpook National University, Daegu, South KoreaDepartment of Information and Communications Engineering, Pukyong National University, Busan, South KoreaDepartment of Electrical and Computer Engineering, Queen's University, Kingston, CanadaTechnology Research Safety Department, KORAIL Research Safety Institute, Daejeon, South KoreaTechnology Research Safety Department, KORAIL Research Safety Institute, Daejeon, South KoreaDepartment of Artificial Intelligence, Kyungpook National University, Daegu, South KoreaDeep learning models learn powerful representational spaces required for handling complex tasks. Recently, data augmentation techniques, region-level, and image-level augmentation have proved effective in significantly improving deep learning models’ generalization performance. Nevertheless, such methods may lose some critical features or are still computationally heavy (or inefficient) due to additional computation burdens. To address this issue, in this paper, we present a novel unified data augmentation method for deep learning models, namely, RandMixAugment, which effectively combines the intrinsic properties of region-level augmentation and image-level augmentation. Specifically, the proposed RandMixAugment employs automated augmentation with masking and mixing operations. Experiments are conducted on well-known CIFAR datasets (CIFAR-10 and CIFAR-100) to verify the effectiveness of the proposed scheme compared to state-of-the-art augmentation techniques. The experimental results demonstrate that the proposed RandMixAugment yields superior performance over state-of-the-art techniques on image classification tasks and further improves the performance of the baseline deep learning model by 1.2% and 2.4% on CIFAR-10 and CIFAR-100 datasets, respectively.https://ieeexplore.ieee.org/document/10217802/Classificationdata augmentationdeep learningimage processingsupervised learning |
| spellingShingle | Yosoeb Shin Vikas Palakonda Sangseok Yun Il-Min Kim Seon-Gon Kim Sang-Mi Park Jae-Mo Kang RandMixAugment: A Novel Unified Technique for Region- and Image-Level Data Augmentations IEEE Access Classification data augmentation deep learning image processing supervised learning |
| title | RandMixAugment: A Novel Unified Technique for Region- and Image-Level Data Augmentations |
| title_full | RandMixAugment: A Novel Unified Technique for Region- and Image-Level Data Augmentations |
| title_fullStr | RandMixAugment: A Novel Unified Technique for Region- and Image-Level Data Augmentations |
| title_full_unstemmed | RandMixAugment: A Novel Unified Technique for Region- and Image-Level Data Augmentations |
| title_short | RandMixAugment: A Novel Unified Technique for Region- and Image-Level Data Augmentations |
| title_sort | randmixaugment a novel unified technique for region and image level data augmentations |
| topic | Classification data augmentation deep learning image processing supervised learning |
| url | https://ieeexplore.ieee.org/document/10217802/ |
| work_keys_str_mv | AT yosoebshin randmixaugmentanovelunifiedtechniqueforregionandimageleveldataaugmentations AT vikaspalakonda randmixaugmentanovelunifiedtechniqueforregionandimageleveldataaugmentations AT sangseokyun randmixaugmentanovelunifiedtechniqueforregionandimageleveldataaugmentations AT ilminkim randmixaugmentanovelunifiedtechniqueforregionandimageleveldataaugmentations AT seongonkim randmixaugmentanovelunifiedtechniqueforregionandimageleveldataaugmentations AT sangmipark randmixaugmentanovelunifiedtechniqueforregionandimageleveldataaugmentations AT jaemokang randmixaugmentanovelunifiedtechniqueforregionandimageleveldataaugmentations |