Data-augmented regression with generative convolutional network
- Publication Type:
- Conference Proceeding
- Citation:
- Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics), 2018, 11234 LNCS pp. 301 - 311
- Issue Date:
- 2018-01-01
Closed Access
| Filename | Description | Size | |||
|---|---|---|---|---|---|
| Ning2018_Chapter_Data-AugmentedRegressionWithGe.pdf | Published version | 1.18 MB |
Copyright Clearance Process
- Recently Added
- In Progress
- Closed Access
This item is closed access and not available.
© Springer Nature Switzerland AG 2018. Generative adversarial networks (GAN)-based approaches have been extensively investigated whereas GAN-inspired regression (i.e., numeric prediction) has rarely been studied in image and video processing domains. The lack of sufficient labeled data in many real-world cases poses great challenges to regression methods, which generally require sufficient labeled samples for their training. In this regard, we propose a unified framework that combines a robust autoencoder and a generative convolutional neural network (GCNN)-based regression model to address the regression problem. Our model is able to generate high-quality artificial samples via augmenting the size of a small number of training samples for better training effects. Extensive experiments are conducted on two real-world datasets and the results show that our proposed model consistently outperforms a set of advanced techniques under various evaluation metrics.
Please use this identifier to cite or link to this item: