Soft Curriculum for Learning Conditional GANs with Noisy-Labeled and Uncurated Unlabeled Data

TL;DR

This paper introduces a soft curriculum learning approach for conditional GANs that effectively handles noisy-labeled and uncurated unlabeled data, improving robustness and performance in generative modeling.

Contribution

The paper proposes a novel soft curriculum learning method that assigns instance-wise weights to training data, enabling robust conditional GAN training with noisy and uncurated data, outperforming existing methods.

Findings

Outperforms existing semi-supervised and label-noise robust methods.

Matches the performance of semi-supervised GANs with less than half the labeled data.

Effective in handling both noisy labels and uncurated unlabeled data.

Abstract

Label-noise or curated unlabeled data is used to compensate for the assumption of clean labeled data in training the conditional generative adversarial network; however, satisfying such an extended assumption is occasionally laborious or impractical. As a step towards generative modeling accessible to everyone, we introduce a novel conditional image generation framework that accepts noisy-labeled and uncurated unlabeled data during training: (i) closed-set and open-set label noise in labeled data and (ii) closed-set and open-set unlabeled data. To combat it, we propose soft curriculum learning, which assigns instance-wise weights for adversarial training while assigning new labels for unlabeled data and correcting wrong labels for labeled data. Unlike popular curriculum learning, which uses a threshold to pick the training samples, our soft curriculum controls the effect of each training instance by using the weights predicted by the auxiliary classifier, resulting in the preservation of useful samples while ignoring harmful ones. Our experiments show that our approach outperforms existing semi-supervised and label-noise robust methods in terms of both quantitative and qualitative performance. In particular, the proposed approach is able to match the performance of (semi-) supervised GANs even with less than half the labeled data.