DuelGAN: A Duel Between Two Discriminators Stabilizes the GAN Training

Wei, Jiaheng; Liu, Minghao; Luo, Jiahao; Zhu, Andrew; Davis, James; Liu, Yang

doi:10.1007/978-3-031-20050-2_18

Cited by 9 publications

(3 citation statements)

References 20 publications

Supporting

Mentioning

Contrasting

Unclassified

Order By: Relevance

“…The FID score of the DGD improves drastically with learning a more complex parameterized distribution, which lends itself naturally to our approach. When using 20 Gaussian components, the FID decreases to 27.25

0.11 (from three computations using different random seeds, see Section 2), positioning itself between the probabilistic autoencoder (28.0) ( Böhm and Seljak 2022 ) and PeerGAN (21.73) ( Wei et al 2022 ).…”

Section: Resultsmentioning

confidence: 99%

The Deep Generative Decoder: MAP estimation of representations improves modelling of single-cell RNA data

Schuster

Krogh

2023

Bioinformatics

View full text Add to dashboard Cite

Motivation Learning low-dimensional representations of single-cell transcriptomics has become instrumental to its downstream analysis. The state of the art is currently represented by neural network models such as variational autoencoders (VAEs) which use a variational approximation of the likelihood for inference. Results We here present the Deep Generative Decoder (DGD), a simple generative model that computes model parameters and representations directly via maximum a posteriori (MAP) estimation. The DGD handles complex parameterized latent distributions naturally unlike VAEs which typically use a fixed Gaussian distribution, because of the complexity of adding other types. We first show its general functionality on a commonly used benchmark set, Fashion-MNIST. Secondly, we apply the model to multiple single-cell data sets. Here the DGD learns low-dimensional, meaningful and well-structured latent representations with sub-clustering beyond the provided labels. The advantages of this approach are its simplicity and its capability to provide representations of much smaller dimensionality than a comparable VAE. Availability and implementation The code is made available in this GitHub repository. scDGD is available as a python package at https://github.com/Center-for-Health-Data-Science/scDGD. Supplementary information Supplementary data are available at Bioinformatics online.

show abstract

Section: Resultsmentioning

confidence: 99%

The Deep Generative Decoder: MAP estimation of representations improves modelling of single-cell RNA data

Schuster

Krogh

2023

Bioinformatics

View full text Add to dashboard Cite

show abstract

“…Although both use the analogy to learning, they are different in their implementation and motivation. In addition to the above methods, there are some approaches that reduce the risk of falling into local modes by building multiple discriminators or multiple generators, such as Dropout-GAN [24], D2GAN [25], GMAN [26] and DuelGAN [27]. These methods can significantly improve the robustness of the model through the collaboration of multiple discriminators or generators, but they exponentially increase the size and training difficulty.…”

Section: Related Workmentioning

confidence: 99%

Soft Generative Adversarial Network: Combating Mode Collapse in Generative Adversarial Network Training via Dynamic Borderline Softening Mechanism

Li,

Tang

2024

Applied Sciences

View full text Add to dashboard Cite

In this paper, we propose the Soft Generative Adversarial Network (SoftGAN), a strategy that utilizes a dynamic borderline softening mechanism to train Generative Adversarial Networks. This mechanism aims to solve the mode collapse problem and enhance the training stability of the generated outputs. Within the SoftGAN, the objective of the discriminator is to learn a fuzzy concept of real data with a soft borderline between real and generated data. This objective is achieved by balancing the principles of maximum concept coverage and maximum expected entropy of fuzzy concepts. During the early training stage of the SoftGAN, the principle of maximum expected entropy of fuzzy concepts guides the learning process due to the significant divergence between the generated and real data. However, in the final stage of training, the principle of maximum concept coverage dominates as the divergence between the two distributions decreases. The dynamic borderline softening mechanism of the SoftGAN can be likened to a student (the generator) striving to create realistic images, with the tutor (the discriminator) dynamically guiding the student towards the right direction and motivating effective learning. The tutor gives appropriate encouragement or requirements according to abilities of the student at different stages, so as to promote the student to improve themselves better. Our approach offers both theoretical and practical benefits for improving GAN training. We empirically demonstrate the superiority of our SoftGAN approach in addressing mode collapse issues and generating high-quality outputs compared to existing approaches.

show abstract

“…На основе 6 можно рассчитать дивергенцию Йенсена-Шеннона между двумя распределениями, используя формулу: Существует несколько разновидностей GAN, которые предназначены для различных приложений и контекстов. Например, при полунаблюдаемом обучении дискриминатор обновляется для присвоения реальных меток классам с 1 по K-1 и поддельной метки классу K, при этом генератор пытается обмануть дискриминатор, чтобы присвоить меньшую метку [16]. В данном исследовании для улучшения обучения DCGAN был использован модифицированный метод одностороннего сглаживания меток.…”

Section: рис 1 архитектура генеративной состязательной сетиunclassified

Exploring the application of neural networks for facial image reconstruction in recognition systems

Markin¹,

Zuparova²,

Martyshkin³

2022

Proceedings of ISP RAS

View full text Add to dashboard Cite

Identifying a person in a digital image using computer vision is a crucial aspect of this field. The presence of external objects, such as medical masks that cover part of the face, can drastically reduce recognition accuracy and increase errors from 5% to 50%, depending on the algorithm. This paper investigates the use of neural networks, in particular the generative adversarial network (GAN), to solve the problem of reconstructing an image of a face covered by a medical mask to improve face recognition accuracy.

show abstract

DuelGAN: A Duel Between Two Discriminators Stabilizes the GAN Training

Cited by 9 publications

References 20 publications

The Deep Generative Decoder: MAP estimation of representations improves modelling of single-cell RNA data

The Deep Generative Decoder: MAP estimation of representations improves modelling of single-cell RNA data

Soft Generative Adversarial Network: Combating Mode Collapse in Generative Adversarial Network Training via Dynamic Borderline Softening Mechanism

Exploring the application of neural networks for facial image reconstruction in recognition systems

Contact Info

Product

Resources

About