Adaptive Feature Interpolation for Low-Shot Image Generation

TL;DR

This paper introduces an unsupervised feature interpolation method for low-shot image generation that stabilizes training and improves sample quality without requiring labels.

Contribution

It presents a novel implicit data augmentation technique using feature space interpolation, enhancing low-shot generative model performance.

Findings

Significantly improves few-shot generation results

Enables stable GAN training with limited data

Achieves high-quality sample synthesis without labels

Abstract

Training of generative models especially Generative Adversarial Networks can easily diverge in low-data setting. To mitigate this issue, we propose a novel implicit data augmentation approach which facilitates stable training and synthesize high-quality samples without need of label information. Specifically, we view the discriminator as a metric embedding of the real data manifold, which offers proper distances between real data points. We then utilize information in the feature space to develop a fully unsupervised and data-driven augmentation method. Experiments on few-shot generation tasks show the proposed method significantly improve results from strong baselines with hundreds of training samples.