Negative Data Augmentation
This work provides a novel weak supervision strategy for a range of unsupervised learning tasks, benefiting researchers and practitioners in generative modeling and representation learning by leveraging knowledge of invalid data.
This paper explores negative data augmentation (NDA) to create out-of-distribution samples, which are then used to improve generative modeling and representation learning. The authors introduce a new GAN objective that incorporates NDA, leading to improved conditional/unconditional image generation and anomaly detection. Additionally, NDA is integrated into a contrastive learning framework, achieving improved performance on downstream tasks like image classification, object detection, and action recognition.
Data augmentation is often used to enlarge datasets with synthetic samples generated in accordance with the underlying data distribution. To enable a wider range of augmentations, we explore negative data augmentation strategies (NDA)that intentionally create out-of-distribution samples. We show that such negative out-of-distribution samples provide information on the support of the data distribution, and can be leveraged for generative modeling and representation learning. We introduce a new GAN training objective where we use NDA as an additional source of synthetic data for the discriminator. We prove that under suitable conditions, optimizing the resulting objective still recovers the true data distribution but can directly bias the generator towards avoiding samples that lack the desired structure. Empirically, models trained with our method achieve improved conditional/unconditional image generation along with improved anomaly detection capabilities. Further, we incorporate the same negative data augmentation strategy in a contrastive learning framework for self-supervised representation learning on images and videos, achieving improved performance on downstream image classification, object detection, and action recognition tasks. These results suggest that prior knowledge on what does not constitute valid data is an effective form of weak supervision across a range of unsupervised learning tasks.