Pioneer Networks: Progressively Growing Generative Autoencoder

Heljakka, Ari; Solin, Arno; Kannala, Juho

doi:10.1007/978-3-030-20887-5_2

Cited by 28 publications

(54 citation statements)

References 8 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…This makes it difficult to train encoder using a pre-trained generator. Another disadvantage of studies in this category is that they are inadequate to generate high-resolution synthetic images; most of these studies cannot go beyond 64x64 resolutions [18]. Moreover, most of these studies fail to extract the latent vector representation of an image that changes only a subset of attributes while preserving the other properties as they are.…”

Section: Related Workmentioning

confidence: 99%

“…The third category is Adversarial Generator-Encoder (AGE) architectures whose structure does not contain a discriminator network and an adversarial game is set up directly between the encoder and the generator [18,19]. In this category, training is performed using a combination of adversarial loss and reconstruction loss that encourages the encoder and generator to be reciprocal.…”

Section: Related Workmentioning

confidence: 99%

“…Most of the models that infer the latent z vectors work with 32x32 resolution images. There are very few models that go beyond 64x64 pixel resolution [18]. Current state-of-the art models in the z inference produce a maximum of 128x128 images.…”

Section: Celeba Modelmentioning

confidence: 99%

“…In Table 2, we compare our generator with the AGE and Pioneer [18] generators for 128x128 pixel resolution, since these models contain a model that infer the latent z vectors similar to ours. Pioneer is the progressive model of the AGE architecture.…”

Section: Fid [38]mentioning

confidence: 99%

See 3 more Smart Citations

Semi-supervised image attribute editing using generative adversarial networks

Dogan

Keleş

2020

Neurocomputing

View full text Add to dashboard Cite

Image attribute editing is a challenging problem that has been recently studied by many researchers using generative networks. The challenge is in the manipulation of selected attributes of images while preserving the other details. The method to achieve this goal is to find an accurate latent vector representation of an image and a direction corresponding to the attribute. Almost all the works in the literature use labeled datasets in a supervised setting for this purpose. In this study, we introduce an architecture called Cyclic Reverse Generator (CRG), which allows learning the inverse function of the generator accurately via an encoder in an unsupervised setting by utilizing cyclic cost minimization. Attribute editing is then performed using the CRG models for finding desired attribute representations in the latent space. In this work, we use two arbitrary reference images, with and without desired attributes, to compute an attribute direction for editing. We show that the proposed approach performs better in terms of image reconstruction compared to the existing end-to-end generative models both quantitatively and qualitatively. We demonstrate state-of-the-art results on both real images and generated images in MNIST and CelebA datasets.

show abstract

Section: Related Workmentioning

confidence: 99%

Section: Related Workmentioning

confidence: 99%

Section: Celeba Modelmentioning

confidence: 99%

Section: Fid [38]mentioning

confidence: 99%

See 2 more Smart Citations

Semi-supervised image attribute editing using generative adversarial networks

Dogan

Keleş

2020

Neurocomputing

View full text Add to dashboard Cite

show abstract

“…The PGAN methods have recently demonstrated a marked improvement in modeling high resolution images as measured by Fréchet Inception Distance (FID) and Inception Score (IS). 10,[18][19][20] In addition, by incrementally growing the GAN the speed of network convergence is increased as more images, through greater batch sizes at low image resolutions, can be presented to the discriminator in less time. We use the official TensorFlow implementation of PGANs Github repository.…”

Section: Semantic Label Modelingmentioning

confidence: 99%

Conditional generative adversarial networks for data augmentation and adaptation in remotely sensed imagery

Howe

Pula

2019

Applications of Machine Learning

View full text Add to dashboard Cite

The difficulty in obtaining labeled data relevant to a given task is among the most common and well-known practical obstacles to applying deep learning techniques to new or even slightly modified domains. The data volumes required by the current generation of supervised learning algorithms typically far exceed what a human needs to learn and complete a given task. We investigate ways to expand a given labeled corpus of remote sensed imagery into a larger corpus using Generative Adversarial Networks (GANs). We then measure how these additional synthetic data affect supervised machine learning performance on an object detection task.Our data driven strategy is to train GANs to (1) generate synthetic segmentation masks and (2) generate plausible synthetic remote sensing imagery corresponding to these segmentation masks. Run sequentially, these GANs allow the generation of synthetic remote sensing imagery complete with segmentation labels. We apply this strategy to the data set from ISPRS' 2D Semantic Labeling Contest -Potsdam, with a follow on vehicle detection task. We find that in scenarios with limited training data, augmenting the available data with such synthetically generated data can improve detector performance.

show abstract

Metrics and methods for robustness evaluation of neural networks with generative models

2021

View full text Add to dashboard Cite

Recent studies have shown that modern deep neural network classifiers are easy to fool, assuming that an adversary is able to slightly modify their inputs. Many papers have proposed adversarial attacks, defenses and methods to measure robustness to such adversarial perturbations. However, most commonly considered adversarial examples are based on perturbations in the input space of the neural network that are unlikely to arise naturally. Recently, especially in computer vision, researchers discovered "natural" perturbations, such as rotations, changes of brightness, or more high-level changes, but these perturbations have not yet been systematically used to measure the performance of classifiers. In this paper, we propose several metrics to measure robustness of classifiers to natural adversarial examples, and methods to evaluate them. These metrics, called latent space performance metrics, are based on the ability of generative models to capture probability distributions. On four image classification case studies, we evaluate the proposed metrics for several classifiers, including ones trained in conventional and robust ways. We find that the latent counterparts of adversarial robustness are associated with the accuracy of the classifier rather than its conventional adversarial robustness, but the latter is still reflected on the properties of found latent perturbations. In addition, our novel method of finding latent adversarial perturbations demonstrates that these perturbations are often perceptually small.

show abstract

Pioneer Networks: Progressively Growing Generative Autoencoder

Cited by 28 publications

References 8 publications

Semi-supervised image attribute editing using generative adversarial networks

Semi-supervised image attribute editing using generative adversarial networks

Conditional generative adversarial networks for data augmentation and adaptation in remotely sensed imagery

Metrics and methods for robustness evaluation of neural networks with generative models

Contact Info

Product

Resources

About