Conditional Wasserstein generative adversarial network-gradient penalty-based approach to alleviating imbalanced data classification

Zheng, Ming; Li, Tong; Zhu, Rui; Tang, Yahui; Tang, Mingjing; Liang, Lin; Ma, Zifei

doi:10.1016/j.ins.2019.10.014

Cited by 128 publications

(53 citation statements)

References 29 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…As a reference, we used a dedicatedly trained, conditional Wasserstein-GAN (cW-GAN) [12,33]. This network used the same architecture and training routine as for the unconstrained GAN above.…”

Section: Conditional Digit Generationmentioning

confidence: 99%

Bayesian Reasoning with Trained Neural Networks

Knollmüller

Enßlin

2021

Entropy

View full text Add to dashboard Cite

We showed how to use trained neural networks to perform Bayesian reasoning in order to solve tasks outside their initial scope. Deep generative models provide prior knowledge, and classification/regression networks impose constraints. The tasks at hand were formulated as Bayesian inference problems, which we approximately solved through variational or sampling techniques. The approach built on top of already trained networks, and the addressable questions grew super-exponentially with the number of available networks. In its simplest form, the approach yielded conditional generative models. However, multiple simultaneous constraints constitute elaborate questions. We compared the approach to specifically trained generators, showed how to solve riddles, and demonstrated its compatibility with state-of-the-art architectures.

show abstract

Section: Conditional Digit Generationmentioning

confidence: 99%

Bayesian Reasoning with Trained Neural Networks

Knollmüller

Enßlin

2021

Entropy

View full text Add to dashboard Cite

show abstract

“…Wasserstein CGAN (WCGAN) is often discussed under WC-GAN as in [63] or CWGAN as in [64], [65]. Similar to CGAN discussed in Section IV-C, the generator and discriminator in WCGAN can be conditioned on the failure class labels y, as auxiliary information.…”

Section: E Wasserstein Cgan (Wcgan)mentioning

confidence: 99%

“…Zheng et al [64] have used WCGAN-GP as an oversampling approach to generate realistic minority samples based on learning the real distribution of available minority samples. In comparison to WGAN-GP, the authors show that incorporating the class label in the WCGAN-GP increases the quality of synthetic generative data.…”

Section: E Wasserstein Cgan (Wcgan)mentioning

confidence: 99%

Learning With Imbalanced Data in Smart Manufacturing: A Comparative Analysis

2021

View full text Add to dashboard Cite

“…For instance, inspired by the idea of AC-GAN (Auxiliary Classifier-GAN) [34], Ali-Gombe and Elyan proposed an improved model MFC-GAN (Multiple Fake Class-GAN) [35] and used the MFC-GAN to handle imbalanced data classification problem. Zheng et al [36] proposed a synthetic oversampling approach for imbalanced data sets.…”

Section: Related Workmentioning

confidence: 99%

Binary Imbalanced Data Classification Based on Modified D2GAN Oversampling and Classifier Fusion

Zhai

Zhang

2020

IEEE Access

View full text Add to dashboard Cite

Binary imbalance problem refers to such a classification scenario where one class contains a large number of samples while another class contains only a few samples. When traditional classifiers face with imbalanced datasets, they usually bias towards majority class resulting in poor classification performance. Oversampling is an effective method to address this problem, yet how to conduct diversity oversampling is a challenge. In this paper, we proposed a diversity oversampling method based on a modified D2GAN model, and on the basis of diversity oversampling, we also proposed a binary imbalanced data classification approach based on classifier fusion by fuzzy integral. Extensive experiments are conducted on 8 data sets to compare the proposed methods with 7 state-of-the-art methods on 5 aspects: MMD-score, Silhouette-score, F-measure, G-means, and AUC-area. The 7 methods include 4 SMOTE related approaches and 3 GAN related approaches. The experimental results demonstrate that the proposed methods are more effective and efficient than the compared approaches. INDEX TERMS Binary class imbalance, diversity oversampling, generative adversarial network, classifier fusion, fuzzy integral

show abstract

Conditional Wasserstein generative adversarial network-gradient penalty-based approach to alleviating imbalanced data classification

Cited by 128 publications

References 29 publications

Bayesian Reasoning with Trained Neural Networks

Bayesian Reasoning with Trained Neural Networks

Learning With Imbalanced Data in Smart Manufacturing: A Comparative Analysis

Binary Imbalanced Data Classification Based on Modified D2GAN Oversampling and Classifier Fusion

Contact Info

Product

Resources

About