GA3N: Generative adversarial AutoAugment network

Vanchinbal Chinbat, Seung-Hwan Bae

Abstract: Highlights•Proposing a new adversarial AutoAugment framework with several GANs.•Integrating two adversarial learning approaches for hard and synthetic samples.•Presenting two different methods for constructing policy search space.•Presenting distinct improvement compared with recent AutoAugment methods.

External IDs:doi:10.1016/j.patcog.2022.108637