Validation Feedback based Image Transfer Network for Data Augmentation

Weili Chen, Seiichiro Kamata, Zitang Sun

研究成果: Conference contribution

抄録

Modern image classifiers are often suffering over-fitting problems because of the insufficient number of images in the dataset. Data augmentation is a strategy to increase the number of training samples. However, recent data augmentation methods are designed manually and cannot generate real-like images. Some neural network-based image generation methods such as GAN and VAE can also be used for data augmentation, but they are usually applied to unbalanced datasets. Since the generated images cannot be guaranteed to be from the same label, using them to extend a balanced dataset may lead to decreasing the accuracy of the classifier. In this paper, we propose an image transfer network to produce images that automatically adapt to a specific dataset and classifier. The image transfer network will search for the output images which can maximize the validation accuracy and help the classifier to overcome the over-fitting problems. Through the experiments, our method achieves high accuracy on CIFAR-10 and CIFAR-100 datasets. Moreover, since it could combine with other data augmentation methods, we show that using our method can push the state-of-the-art results furthermore.

本文言語English
ホスト出版物のタイトルProceedings of 2020 2nd International Conference on Video, Signal and Image Processing, VSIP 2020
出版社Association for Computing Machinery
ページ23-29
ページ数7
ISBN(電子版)9781450388931
DOI
出版ステータスPublished - 2020 4 12
イベント2nd International Conference on Video, Signal and Image Processing, VSIP 2020 - Virtual, Online, Indonesia
継続期間: 2020 12 42020 12 6

出版物シリーズ

名前ACM International Conference Proceeding Series

Conference

Conference2nd International Conference on Video, Signal and Image Processing, VSIP 2020
国/地域Indonesia
CityVirtual, Online
Period20/12/420/12/6

ASJC Scopus subject areas

  • ソフトウェア
  • 人間とコンピュータの相互作用
  • コンピュータ ビジョンおよびパターン認識
  • コンピュータ ネットワークおよび通信

フィンガープリント

「Validation Feedback based Image Transfer Network for Data Augmentation」の研究トピックを掘り下げます。これらがまとまってユニークなフィンガープリントを構成します。

引用スタイル