UGA-GAN: Unified Geometry-Aware GAN for Enhanced Training and Generation of High-Dimensional Data

<p dir="ltr">Generative Adversarial Networks (GANs) have shown impressive performance in generating realistic data across various domains. However, they suffer from key challenges such as mode collapse, latent space disorganization, and geometric inconsistency, which hinder the gener...

وصف كامل

محفوظ في:
التفاصيل البيبلوغرافية
المؤلف الرئيسي: Wasi Ahmad (23124064) (author)
مؤلفون آخرون: Md. Faysal Ahamed (21842396) (author), Amith Khandakar (14151981) (author), SM Ashfaq Uz Zaman (23124067) (author), Mohamed Arselene Ayari (17873878) (author)
منشور في: 2025
الموضوعات:
الوسوم: إضافة وسم
لا توجد وسوم, كن أول من يضع وسما على هذه التسجيلة!
_version_ 1864513523989610496
author Wasi Ahmad (23124064)
author2 Md. Faysal Ahamed (21842396)
Amith Khandakar (14151981)
SM Ashfaq Uz Zaman (23124067)
Mohamed Arselene Ayari (17873878)
author2_role author
author
author
author
author_facet Wasi Ahmad (23124064)
Md. Faysal Ahamed (21842396)
Amith Khandakar (14151981)
SM Ashfaq Uz Zaman (23124067)
Mohamed Arselene Ayari (17873878)
author_role author
dc.creator.none.fl_str_mv Wasi Ahmad (23124064)
Md. Faysal Ahamed (21842396)
Amith Khandakar (14151981)
SM Ashfaq Uz Zaman (23124067)
Mohamed Arselene Ayari (17873878)
dc.date.none.fl_str_mv 2025-10-23T15:00:00Z
dc.identifier.none.fl_str_mv 10.1109/access.2025.3621108
dc.relation.none.fl_str_mv https://figshare.com/articles/journal_contribution/UGA-GAN_Unified_Geometry-Aware_GAN_for_Enhanced_Training_and_Generation_of_High-Dimensional_Data/31239316
dc.rights.none.fl_str_mv CC BY 4.0
info:eu-repo/semantics/openAccess
dc.subject.none.fl_str_mv Information and computing sciences
Artificial intelligence
Machine learning
Manifold learning
latent space regularization
mode collapse
geometry aware GAN
Generative adversarial networks
Training
Generators
Videos
Geometry
High dimensional data
Shape
Data collection
dc.title.none.fl_str_mv UGA-GAN: Unified Geometry-Aware GAN for Enhanced Training and Generation of High-Dimensional Data
dc.type.none.fl_str_mv Text
Journal contribution
info:eu-repo/semantics/publishedVersion
text
contribution to journal
description <p dir="ltr">Generative Adversarial Networks (GANs) have shown impressive performance in generating realistic data across various domains. However, they suffer from key challenges such as mode collapse, latent space disorganization, and geometric inconsistency, which hinder the generation of high-fidelity and diverse samples in complex, high-dimensional datasets. In this paper, we introduce a novel framework, the Unified Geometry-Aware Generative Adversarial Network (UGA-GAN), which integrates manifold learning, latent space regularization, and geometry-aware discrimination to address these challenges. We propose a unified architecture that enhances GAN performance by aligning generated samples with the underlying data manifold, promoting smooth and interpretable latent space representations, and ensuring global geometric consistency. Our experimental evaluation on the CIFAR-10 dataset demonstrates that UGA-GAN outperforms several baseline models, including DCGAN, WGAN, and LSGAN, achieving a 37.97% reduction in Fréchet Inception Distance (FID) score, indicating superior sample quality and diversity. Furthermore, t-SNE visualizations confirm that UGA-GAN generates more coherent and diverse samples with better mode coverage. These results highlight the potential of UGA-GAN to significantly improve high-dimensional data generation tasks in domains such as medical imaging, video generation, and 3D object synthesis. While UGA-GAN presents state-of-the-art performance, future work will focus on optimizing its computational efficiency, scaling it to larger datasets, and integrating it with emerging models such as diffusion networks and reinforcement learning for further performance enhancement.</p><h2 dir="ltr">Other Information</h2><p dir="ltr">Published in: IEEE Access<br>License: <a href="https://creativecommons.org/licenses/by/4.0/deed.en" target="_blank">https://creativecommons.org/licenses/by/4.0/</a><br>See article on publisher's website: <a href="https://dx.doi.org/10.1109/access.2025.3621108" target="_blank">https://dx.doi.org/10.1109/access.2025.3621108</a></p>
eu_rights_str_mv openAccess
id Manara2_f5aaa39602484144c70d39a6b4291c06
identifier_str_mv 10.1109/access.2025.3621108
network_acronym_str Manara2
network_name_str Manara2
oai_identifier_str oai:figshare.com:article/31239316
publishDate 2025
repository.mail.fl_str_mv
repository.name.fl_str_mv
repository_id_str
rights_invalid_str_mv CC BY 4.0
spelling UGA-GAN: Unified Geometry-Aware GAN for Enhanced Training and Generation of High-Dimensional DataWasi Ahmad (23124064)Md. Faysal Ahamed (21842396)Amith Khandakar (14151981)SM Ashfaq Uz Zaman (23124067)Mohamed Arselene Ayari (17873878)Information and computing sciencesArtificial intelligenceMachine learningManifold learninglatent space regularizationmode collapsegeometry aware GANGenerative adversarial networksTrainingGeneratorsVideosGeometryHigh dimensional dataShapeData collection<p dir="ltr">Generative Adversarial Networks (GANs) have shown impressive performance in generating realistic data across various domains. However, they suffer from key challenges such as mode collapse, latent space disorganization, and geometric inconsistency, which hinder the generation of high-fidelity and diverse samples in complex, high-dimensional datasets. In this paper, we introduce a novel framework, the Unified Geometry-Aware Generative Adversarial Network (UGA-GAN), which integrates manifold learning, latent space regularization, and geometry-aware discrimination to address these challenges. We propose a unified architecture that enhances GAN performance by aligning generated samples with the underlying data manifold, promoting smooth and interpretable latent space representations, and ensuring global geometric consistency. Our experimental evaluation on the CIFAR-10 dataset demonstrates that UGA-GAN outperforms several baseline models, including DCGAN, WGAN, and LSGAN, achieving a 37.97% reduction in Fréchet Inception Distance (FID) score, indicating superior sample quality and diversity. Furthermore, t-SNE visualizations confirm that UGA-GAN generates more coherent and diverse samples with better mode coverage. These results highlight the potential of UGA-GAN to significantly improve high-dimensional data generation tasks in domains such as medical imaging, video generation, and 3D object synthesis. While UGA-GAN presents state-of-the-art performance, future work will focus on optimizing its computational efficiency, scaling it to larger datasets, and integrating it with emerging models such as diffusion networks and reinforcement learning for further performance enhancement.</p><h2 dir="ltr">Other Information</h2><p dir="ltr">Published in: IEEE Access<br>License: <a href="https://creativecommons.org/licenses/by/4.0/deed.en" target="_blank">https://creativecommons.org/licenses/by/4.0/</a><br>See article on publisher's website: <a href="https://dx.doi.org/10.1109/access.2025.3621108" target="_blank">https://dx.doi.org/10.1109/access.2025.3621108</a></p>2025-10-23T15:00:00ZTextJournal contributioninfo:eu-repo/semantics/publishedVersiontextcontribution to journal10.1109/access.2025.3621108https://figshare.com/articles/journal_contribution/UGA-GAN_Unified_Geometry-Aware_GAN_for_Enhanced_Training_and_Generation_of_High-Dimensional_Data/31239316CC BY 4.0info:eu-repo/semantics/openAccessoai:figshare.com:article/312393162025-10-23T15:00:00Z
spellingShingle UGA-GAN: Unified Geometry-Aware GAN for Enhanced Training and Generation of High-Dimensional Data
Wasi Ahmad (23124064)
Information and computing sciences
Artificial intelligence
Machine learning
Manifold learning
latent space regularization
mode collapse
geometry aware GAN
Generative adversarial networks
Training
Generators
Videos
Geometry
High dimensional data
Shape
Data collection
status_str publishedVersion
title UGA-GAN: Unified Geometry-Aware GAN for Enhanced Training and Generation of High-Dimensional Data
title_full UGA-GAN: Unified Geometry-Aware GAN for Enhanced Training and Generation of High-Dimensional Data
title_fullStr UGA-GAN: Unified Geometry-Aware GAN for Enhanced Training and Generation of High-Dimensional Data
title_full_unstemmed UGA-GAN: Unified Geometry-Aware GAN for Enhanced Training and Generation of High-Dimensional Data
title_short UGA-GAN: Unified Geometry-Aware GAN for Enhanced Training and Generation of High-Dimensional Data
title_sort UGA-GAN: Unified Geometry-Aware GAN for Enhanced Training and Generation of High-Dimensional Data
topic Information and computing sciences
Artificial intelligence
Machine learning
Manifold learning
latent space regularization
mode collapse
geometry aware GAN
Generative adversarial networks
Training
Generators
Videos
Geometry
High dimensional data
Shape
Data collection