Subsampling Generative Adversarial Networks: Density Ratio Estimation in Feature Space With Softplus Loss

Ding, Xin; Wang, Z. Jane; Welch, William J.

doi:10.1109/tsp.2020.2979601

Cited by 10 publications

(4 citation statements)

References 26 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…DRS improves the quality of generated samples by filtering bad-quality generation results. Ding et al (Ding, Wang, and Welch 2020) designed a novel Softplus loss for discriminator-based density ratio estimation. The algorithm does not depend on the optimization of discriminator and is applicable to multiple GANs.…”

Section: Model Perception Rectification Algorithms Based On Model-rep...mentioning

confidence: 99%

Generative Model Perception Rectification Algorithm for Trade-Off between Diversity and Quality

Lan,

Xiao,

Yang

et al. 2024

AAAI

View full text Add to dashboard Cite

How to balance the diversity and quality of results from generative models through perception rectification poses a significant challenge. Abnormal perception in generative models is typically caused by two factors: inadequate model structure and imbalanced data distribution. In response to this issue, we propose the dynamic model perception rectification algorithm (DMPRA) for generalized generative models. The core idea is to gain a comprehensive perception of the data in the generative model by appropriately highlighting the low-density samples in the perception space, also known as the minor group samples. The entire process can be summarized as "search-evaluation-adjustment". To identify low-density regions in the data manifold within the perception space of generative models, we introduce a filtering method based on extended neighborhood sampling. Based on the informational value of samples from low-density regions, our proposed mechanism generates informative weights to assess the significance of these samples in correcting the models' perception. By using dynamic adjustment, DMPRA ensures simultaneous enhancement of diversity and quality in the presence of imbalanced data distribution. Experimental results indicate that the algorithm has effectively improved Generative Adversarial Nets (GANs), Normalizing Flows (Flows), Variational Auto-Encoders (VAEs), and Diffusion Models (Diffusion).

show abstract

Section: Model Perception Rectification Algorithms Based On Model-rep...mentioning

confidence: 99%

Generative Model Perception Rectification Algorithm for Trade-Off between Diversity and Quality

Lan,

Xiao,

Yang

et al. 2024

AAAI

View full text Add to dashboard Cite

show abstract

“…The classical FID is simply defined using the marginals of y and ŷ and will be denoted as MFID in the rest of the paper, to emphasize that it a distance between the marginal distributions. After some algebraic manipulations, plugging these into MWD yields [10], [30]…”

Section: A Derivationmentioning

confidence: 99%

Conditional Frechet Inception Distance

Soloveitchik¹,

Diskin²,

Morin³

et al. 2021

Preprint

View full text Add to dashboard Cite

We consider distance functions between conditional distributions functions. We focus on the Wasserstein metric and its Gaussian case known as the Frechet Inception Distance (FID). We develop conditional versions of these metrics, and analyze their relations. Then, we numerically compare the metrics in the context of performance evaluation of conditional generative models. Our results show that the metrics are similar in classical models which are less susceptible to conditional collapse. But the conditional distances are more informative in modern unsupervised, semisupervised and unpaired models where learning the relations between the inputs and outputs is the main challenge.

show abstract

“…Therefore, directly using the generator's distribution 𝑝 𝑔 does not guarantee the high fidelity and it only minimizes the distribution discrepancy between the attack model and the target model. We explain this by a simple example on Figure 5, which is popular in the GAN literature [3,15,64].…”

Section: Motivation and Problem Formulationmentioning

confidence: 99%

Model Extraction and Defenses on Generative Adversarial Networks

Hu,

Pang

2021

Preprint

View full text Add to dashboard Cite

Model extraction attacks aim to duplicate a machine learning model through query access to a target model. Early studies mainly focus on discriminative models. Despite the success, model extraction attacks against generative models are less well explored. In this paper, we systematically study the feasibility of model extraction attacks against generative adversarial networks (GANs). Specifically, we first define accuracy and fidelity on model extraction attacks against GANs. Then we study model extraction attacks against GANs from the perspective of accuracy extraction and fidelity extraction, according to the adversary's goals and background knowledge. We further conduct a case study where an adversary can transfer knowledge of the extracted model which steals a stateof-the-art GAN trained with more than 3 million images to new domains to broaden the scope of applications of model extraction attacks. Finally, we propose effective defense techniques to safeguard GANs, considering a trade-off between the utility and security of GAN models. CCS CONCEPTS• Security and privacy; • Computing methodologies → Machine learning;

show abstract

Subsampling Generative Adversarial Networks: Density Ratio Estimation in Feature Space With Softplus Loss

Cited by 10 publications

References 26 publications

Generative Model Perception Rectification Algorithm for Trade-Off between Diversity and Quality

Generative Model Perception Rectification Algorithm for Trade-Off between Diversity and Quality

Conditional Frechet Inception Distance

Model Extraction and Defenses on Generative Adversarial Networks

Contact Info

Product

Resources

About