Differentiable Augmentation for Data-Efficient GAN Training

Zhao, Shengyu; Liu, Zhijian; Lin, Ji; Zhu, Jun‐Yan; Han, Song

doi:10.48550/arxiv.2006.10738

Cited by 63 publications

(123 citation statements)

References 0 publications

Supporting

Mentioning

119

Contrasting

Unclassified

Order By: Relevance

“…where I is a random perturbation of the input image I drawn from a distribution π(•|I) of candidate data augmentations. In our work, we adopt the various data augmentation techniques considered in DiffAugment [38], including random colorization, random translation, random resize, and random cutout.…”

Section: Augclip: Avoiding Adversarial Generationmentioning

confidence: 99%

FuseDream: Training-Free Text-to-Image Generation with Improved CLIP+GAN Space Optimization

Liu¹,

Gong²,

Wu³

et al. 2021

Preprint

View full text Add to dashboard Cite

show abstract

Section: Augclip: Avoiding Adversarial Generationmentioning

confidence: 99%

FuseDream: Training-Free Text-to-Image Generation with Improved CLIP+GAN Space Optimization

Liu¹,

Gong²,

Wu³

et al. 2021

Preprint

View full text Add to dashboard Cite

show abstract

“…Few-shot learning is appealing, but very challenging in training GANs, since data augmentation methods that are developed for discriminative learning tasks are not directly applicable. To address this challenge, differentiable data augmentation methods and variants [63,23,52,64] have been proposed in training GANs with very exciting results obtained. Very recently, equiped with the differentiable data augmentation method [63], a FastGAN approach [35] is proposed to realize light-weight yet sufficiently powerful GANs with several novel designs including the SLE module.…”

Section: Related Work and Our Contributionsmentioning

confidence: 99%

“…To address this challenge, differentiable data augmentation methods and variants [63,23,52,64] have been proposed in training GANs with very exciting results obtained. Very recently, equiped with the differentiable data augmentation method [63], a FastGAN approach [35] is proposed to realize light-weight yet sufficiently powerful GANs with several novel designs including the SLE module. The proposed SLIM is built on the SLE in FastGANs by exploiting the well-known Google's Inception [50,51] building block design and the Átrous convolution [7].…”

Section: Related Work and Our Contributionsmentioning

confidence: 99%

“…Although powerful, as the resolution of synthesized images goes higher, the width and the depth of a generator network goes wider and deeper accordingly, leading to much increased memory footprint and longer training time. Recently, by exploiting the differentiable data augmentation methods [63], Liu et al [35] present a FastGAN approach which further introduces a skip-layer channel-wise excitation (SLE) module as a drop-in component for both generator and discriminator networks, as well as a decoder enhanced design for self-supervised discriminator networks. FastGANs have shown exciting results which outperform the well-known and powerful StyleGANv2 [26] under the low-shot training settings.…”

Section: Introduction 1motivation and Objectivementioning

confidence: 99%

See 1 more Smart Citation

Learning Inception Attention for Image Synthesis and Image Recognition

Shen¹,

Wu²

2021

Preprint

View full text Add to dashboard Cite

Image synthesis and image recognition have witnessed remarkable progress, but often at the expense of computationally expensive training and inference. Learning lightweight yet expressive deep model has emerged as an important and interesting direction. Inspired by the well-known split-transform-aggregate design heuristic in the Inception building block, this paper proposes a Skip-Layer Inception Module (SLIM) that facilitates efficient learning of image synthesis models, and a same-layer variant (dubbed as SLIM too) as a stronger alternative to the well-known ResNeXts for image recognition. The proposed method is built on the idea of developing spatially-adaptive feature modulation/attention in a network. Specifically, SLIM is done from an input source feature map to a target one, either skip-layer or same-layer. In SLIM, the input feature map is first split into a number of groups (e.g., 4). Each group is then transformed to a latent style vector (via channel-wise attention) and a latent spatial mask (via spatial attention). The learned latent masks and latent style vectors are aggregated to modulate the target feature map. For generative learning, SLIM is built on a recently proposed lightweight Generative Adversarial Networks (i.e., FastGANs) which present a skip-layer excitation (SLE) module. For few-shot image synthesis tasks, the proposed SLIM achieves better performance than the SLE work and other related methods. For one-shot image synthesis tasks, it shows stronger capability of preserving image structures than prior arts such as the SinGANs. For image classification tasks, the proposed SLIM is used as a drop-in replacement for convolution layers in ResNets (resulting the ResNeXt-like models) and achieves better accuracy in the ImageNet-1000 dataset, with significantly smaller model complexity.

show abstract

“…Although the missing node generator is able to generate missing nodes for each local graph, it is well known that GAN may perform poorly on small data [34]. Similarly, training an effective GCN model for node classification also requires a large amount of samples.…”

Section: Federated Learningmentioning

confidence: 99%

FedNI: Federated Graph Learning with Network Inpainting for Population-Based Disease Prediction

Peng,

Wang,

Dvornek

et al. 2021

Preprint

View full text Add to dashboard Cite

Graph Convolutional Neural Networks (GCNs) are widely used for graph analysis. Specifically, in medical applications, GCNs can be used for disease prediction on a population graph, where graph nodes represent individuals and edges represent individual similarities. However, GCNs rely on a vast amount of data, which is challenging to collect for a single medical institution. In addition, a critical challenge that most medical institutions continue to face is addressing disease prediction in isolation with incomplete data information. To address these issues, Federated Learning (FL) allows isolated local institutions to collaboratively train a global model without data sharing. In this work, we propose a framework, FedNI, to leverage network inpainting and inter-institutional data via FL. Specifically, we first federatively train missing node and edge predictor using a graph generative adversarial network (GAN) to complete the missing information of local networks. Then we train a global GCN node classifier across institutions using a federated graph learning platform. The novel design enables us to build more accurate machine learning models by leveraging federated learning and also graph learning approaches. We demonstrate that our federated model outperforms local and baseline FL methods with significant margins on two public neuroimaging datasets.

show abstract

Differentiable Augmentation for Data-Efficient GAN Training

Cited by 63 publications

References 0 publications

FuseDream: Training-Free Text-to-Image Generation with Improved CLIP+GAN Space Optimization

FuseDream: Training-Free Text-to-Image Generation with Improved CLIP+GAN Space Optimization

Learning Inception Attention for Image Synthesis and Image Recognition

FedNI: Federated Graph Learning with Network Inpainting for Population-Based Disease Prediction

Contact Info

Product

Resources

About