An Improved Self-supervised GAN via Adversarial Training

Tran, Ngoc-Trung; Tran, Viet-Hung; Nguyen, Nhu; Cheung, Ngai-Man

doi:10.48550/arxiv.1905.05469

Cited by 4 publications

(4 citation statements)

References 22 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…[5] introduced an auxiliary rotation detection task onto the discriminator to alleviate the forgetting issue. [45,46] analyzed the drawback of the self-supervised task in [5] and proposed a multi-class minimax selfsupervised task. [26] combined self-and semi-supervised learning to outperform supervised GANs.…”

Section: Self-supervised Learning In Gansmentioning

confidence: 99%

Self-Supervised GANs with Label Augmentation

Hou¹,

Shen²,

Cao³

et al. 2021

Preprint

View full text Add to dashboard Cite

Recently, transformation-based self-supervised learning has been applied to generative adversarial networks (GANs) to mitigate the catastrophic forgetting problem of discriminator by learning stable representations. However, the separate selfsupervised tasks in existing self-supervised GANs cause an inconsistent goal with generative modeling due to the learning of the generator from their generator distribution-agnostic classifiers. To address this issue, we propose a novel selfsupervised GANs framework with label augmentation, i.e., augmenting the GAN labels (real or fake) with the self-supervised pseudo-labels. In particular, the discriminator and the self-supervised classifier are unified to learn a single task that predicts the augmented label such that the discriminator/classifier is aware of the generator distribution, while the generator tries to confuse the discriminator/classifier by optimizing the discrepancy between the transformed real and generated distributions. Theoretically, we prove that the generator, at the equilibrium point, converges to replicate the data distribution. Empirically, we demonstrate that the proposed method significantly outperforms competitive baselines on both generative modeling and representation learning across benchmark datasets.Preprint. Under review.

show abstract

Section: Self-supervised Learning In Gansmentioning

confidence: 99%

Self-Supervised GANs with Label Augmentation

Hou¹,

Shen²,

Cao³

et al. 2021

Preprint

View full text Add to dashboard Cite

show abstract

“…A number of approaches based on SS have been proven successful for GANs training. Tran et al [24] aim to improve GANs by applying SS learning via the geometric transformation on input images and assign the pseudo-labels to these transformed images. Chen et al propose the selfsupervised GAN [1], adding auxiliary rotation loss as a selfsupervised loss to the discriminator.…”

Section: B Self-supervised Learningmentioning

confidence: 99%

PREGAN: Pose Randomization and Estimation for Weakly Paired Image Style Translation

Chen,

Guo,

et al. 2020

Preprint

View full text Add to dashboard Cite

Utilizing the trained model under different conditions without data annotation is attractive for robot applications. Towards this goal, one class of methods is to translate the image style from the training environment to the current one. Conventional studies on image style translation mainly focus on two settings: paired data on images from two domains with exactly aligned content, and unpaired data, with independent content. In this paper, we would like to propose a new setting, where the content in the two images is aligned with error in poses. We consider that this setting is more practical since robots with various sensors are able to align the data up to some error level, even with different styles. To solve this problem, we propose PRoGAN to learn a style translator by intentionally transforming the original domain images with a noisy pose, then matching the distribution of translated transformed images and the distribution of the target domain images. The adversarial training enforces the network to learn the style translation, avoiding being entangled with other variations. In addition, we propose two pose estimation based self-supervised tasks to further improve the performance. Finally, PRoGAN is validated on both simulated and real-world collected data to show the effectiveness. Results on down-stream tasks, classification, road segmentation, object detection, and feature matching show its potential for real applications. Code is available here.

show abstract

“…Finally, GANs face the perennial problem of mode collapse, where p g collapses to only cover a few modes of p r , resulting in generated samples of limited diversity. Consequently, recent years have seen efforts [4]- [10] to mitigate these GAN problems, including using gradient matching [5] and a two time-scale update rule [7].…”

Section: Introductionmentioning

confidence: 99%

InfoMax-GAN: Improved Adversarial Image Generation via Information Maximization and Contrastive Learning

Lee¹,

Tran²,

Cheung³

2020

Preprint

Self Cite

View full text Add to dashboard Cite

While Generative Adversarial Networks (GANs) are fundamental to many generative modelling applications, they suffer from numerous issues. In this work, we propose a principled framework to simultaneously address two fundamental issues in GANs: catastrophic forgetting of the discriminator and mode collapse of the generator. We achieve this by employing for GANs a contrastive learning and mutual information maximization approach, and perform extensive analyses to understand sources of improvements. Our approach significantly stabilizes GAN training and improves GAN performance for image synthesis across five datasets under the same training and evaluation conditions against state-of-the-art works. Our approach is simple to implement and practical: it involves only one auxiliary objective, has low computational cost, and performs robustly across a wide range of training settings and datasets without any hyperparameter tuning. For reproducibility, our code is available at https://github.com/kwotsin/mimicry.

show abstract

An Improved Self-supervised GAN via Adversarial Training

Cited by 4 publications

References 22 publications

Self-Supervised GANs with Label Augmentation

Self-Supervised GANs with Label Augmentation

PREGAN: Pose Randomization and Estimation for Weakly Paired Image Style Translation

InfoMax-GAN: Improved Adversarial Image Generation via Information Maximization and Contrastive Learning

Contact Info

Product

Resources

About