Exploring Bias in GAN-based Data Augmentation for Small Samples
This addresses bias concerns in GAN-based data augmentation for small-sample machine learning tasks, but appears incremental as it builds on existing validation of GAN-based DA viability.
The paper investigates bias in GAN-based data augmentation for small samples, finding that low bias does not harm performance, and proposes a pipeline to assess dataset augmentability and advice to mitigate bias.
For machine learning task, lacking sufficient samples mean the trained model has low confidence to approach the ground truth function. Until recently, after the generative adversarial networks (GAN) had been proposed, we see the hope of small samples data augmentation (DA) with realistic fake data, and many works validated the viability of GAN-based DA. Although most of the works pointed out higher accuracy can be achieved using GAN-based DA, some researchers stressed that the fake data generated from GAN has inherent bias, and in this paper, we explored when the bias is so low that it cannot hurt the performance, we set experiments to depict the bias in different GAN-based DA setting, and from the results, we design a pipeline to inspect specific dataset is efficiently-augmentable with GAN-based DA or not. And finally, depending on our trial to reduce the bias, we proposed some advice to mitigate bias in GAN-based DA application.