Unsupervised feature learning-based encoder and adversarial networks
Abstract In this paper, we propose a novel deep learning-based feature learning architecture for object classification. Conventionally, deep learning methods are trained with supervised learning for object classification. But, this would require large amount of training data. Currently there are inc...
| Main Authors: | , , , , , , , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
SpringerOpen
2021-09-01
|
| Series: | Journal of Big Data |
| Subjects: | |
| Online Access: | https://doi.org/10.1186/s40537-021-00508-9 |
| _version_ | 1818591964577136640 |
|---|---|
| author | Endang Suryawati Hilman F. Pardede Vicky Zilvan Ade Ramdan Dikdik Krisnandi Ana Heryana R. Sandra Yuwana R. Budiarianto Suryo Kusumo Andria Arisal Ahmad Afif Supianto |
| author_facet | Endang Suryawati Hilman F. Pardede Vicky Zilvan Ade Ramdan Dikdik Krisnandi Ana Heryana R. Sandra Yuwana R. Budiarianto Suryo Kusumo Andria Arisal Ahmad Afif Supianto |
| author_sort | Endang Suryawati |
| collection | DOAJ |
| description | Abstract In this paper, we propose a novel deep learning-based feature learning architecture for object classification. Conventionally, deep learning methods are trained with supervised learning for object classification. But, this would require large amount of training data. Currently there are increasing trends to employ unsupervised learning for deep learning. By doing so, dependency on the availability of large training data could be reduced. One implementation of unsupervised deep learning is for feature learning where the network is designed to “learn” features automatically from data to obtain good representation that then could be used for classification. Autoencoder and generative adversarial networks (GAN) are examples of unsupervised deep learning methods. For GAN however, the trajectories of feature learning may go to unpredicted directions due to random initialization, making it unsuitable for feature learning. To overcome this, a hybrid of encoder and deep convolutional generative adversarial network (DCGAN) architectures, a variant of GAN, are proposed. Encoder is put on top of the Generator networks of GAN to avoid random initialisation. We called our method as EGAN. The output of EGAN is used as features for two deep convolutional neural networks (DCNNs): AlexNet and DenseNet. We evaluate the proposed methods on three types of dataset and the results indicate that better performances are achieved by our proposed method compared to using autoencoder and GAN. |
| first_indexed | 2024-12-16T10:20:48Z |
| format | Article |
| id | doaj.art-4f0f66f0b2214ca9ad9196086b4dbb23 |
| institution | Directory Open Access Journal |
| issn | 2196-1115 |
| language | English |
| last_indexed | 2024-12-16T10:20:48Z |
| publishDate | 2021-09-01 |
| publisher | SpringerOpen |
| record_format | Article |
| series | Journal of Big Data |
| spelling | doaj.art-4f0f66f0b2214ca9ad9196086b4dbb232022-12-21T22:35:18ZengSpringerOpenJournal of Big Data2196-11152021-09-018111710.1186/s40537-021-00508-9Unsupervised feature learning-based encoder and adversarial networksEndang Suryawati0Hilman F. Pardede1Vicky Zilvan2Ade Ramdan3Dikdik Krisnandi4Ana Heryana5R. Sandra Yuwana6R. Budiarianto Suryo Kusumo7Andria Arisal8Ahmad Afif Supianto9Research Center for Informatics, Indonesian Institute of SciencesResearch Center for Informatics, Indonesian Institute of SciencesResearch Center for Informatics, Indonesian Institute of SciencesResearch Center for Informatics, Indonesian Institute of SciencesResearch Center for Informatics, Indonesian Institute of SciencesResearch Center for Informatics, Indonesian Institute of SciencesResearch Center for Informatics, Indonesian Institute of SciencesResearch Center for Informatics, Indonesian Institute of SciencesResearch Center for Informatics, Indonesian Institute of SciencesResearch Center for Informatics, Indonesian Institute of SciencesAbstract In this paper, we propose a novel deep learning-based feature learning architecture for object classification. Conventionally, deep learning methods are trained with supervised learning for object classification. But, this would require large amount of training data. Currently there are increasing trends to employ unsupervised learning for deep learning. By doing so, dependency on the availability of large training data could be reduced. One implementation of unsupervised deep learning is for feature learning where the network is designed to “learn” features automatically from data to obtain good representation that then could be used for classification. Autoencoder and generative adversarial networks (GAN) are examples of unsupervised deep learning methods. For GAN however, the trajectories of feature learning may go to unpredicted directions due to random initialization, making it unsuitable for feature learning. To overcome this, a hybrid of encoder and deep convolutional generative adversarial network (DCGAN) architectures, a variant of GAN, are proposed. Encoder is put on top of the Generator networks of GAN to avoid random initialisation. We called our method as EGAN. The output of EGAN is used as features for two deep convolutional neural networks (DCNNs): AlexNet and DenseNet. We evaluate the proposed methods on three types of dataset and the results indicate that better performances are achieved by our proposed method compared to using autoencoder and GAN.https://doi.org/10.1186/s40537-021-00508-9Generative adversarial networkUnsupervised feature learningAutoencoderConvolutional neural networks |
| spellingShingle | Endang Suryawati Hilman F. Pardede Vicky Zilvan Ade Ramdan Dikdik Krisnandi Ana Heryana R. Sandra Yuwana R. Budiarianto Suryo Kusumo Andria Arisal Ahmad Afif Supianto Unsupervised feature learning-based encoder and adversarial networks Journal of Big Data Generative adversarial network Unsupervised feature learning Autoencoder Convolutional neural networks |
| title | Unsupervised feature learning-based encoder and adversarial networks |
| title_full | Unsupervised feature learning-based encoder and adversarial networks |
| title_fullStr | Unsupervised feature learning-based encoder and adversarial networks |
| title_full_unstemmed | Unsupervised feature learning-based encoder and adversarial networks |
| title_short | Unsupervised feature learning-based encoder and adversarial networks |
| title_sort | unsupervised feature learning based encoder and adversarial networks |
| topic | Generative adversarial network Unsupervised feature learning Autoencoder Convolutional neural networks |
| url | https://doi.org/10.1186/s40537-021-00508-9 |
| work_keys_str_mv | AT endangsuryawati unsupervisedfeaturelearningbasedencoderandadversarialnetworks AT hilmanfpardede unsupervisedfeaturelearningbasedencoderandadversarialnetworks AT vickyzilvan unsupervisedfeaturelearningbasedencoderandadversarialnetworks AT aderamdan unsupervisedfeaturelearningbasedencoderandadversarialnetworks AT dikdikkrisnandi unsupervisedfeaturelearningbasedencoderandadversarialnetworks AT anaheryana unsupervisedfeaturelearningbasedencoderandadversarialnetworks AT rsandrayuwana unsupervisedfeaturelearningbasedencoderandadversarialnetworks AT rbudiariantosuryokusumo unsupervisedfeaturelearningbasedencoderandadversarialnetworks AT andriaarisal unsupervisedfeaturelearningbasedencoderandadversarialnetworks AT ahmadafifsupianto unsupervisedfeaturelearningbasedencoderandadversarialnetworks |