Filters








12,902 Hits in 3.9 sec

Conditional Hybrid GAN for Sequence Generation [article]

Yi Yu, Abhishek Srivastava, Rajiv Ratn Shah
2020 arXiv   pre-print
In this paper, we propose a novel conditional hybrid GAN (C-Hybrid-GAN) to solve this issue. Discrete sequence with triplet attributes are separately generated when conditioned on the same context.  ...  evaluating the task of generating melody (associated with note, duration, and rest) from lyrics, we demonstrate that the proposed C-Hybrid-GAN outperforms the existing methods in context-conditioned discrete-valued  ...  Conditional hybrid GAN We propose an end-to-end deep generative model for generating sequence conditioned on the context.  ... 
arXiv:2009.08616v1 fatcat:lg57jaoavzb2hkvuqqpaqlewg4

The Angel is in the Priors: Improving GAN based Image and Sequence Inpainting with Better Noise and Structural Priors [article]

Avisek Lahiri, Arnav Kumar Jain, Prabir Kumar Biswas
2019 arXiv   pre-print
However, there is a dearth of literature for a fully unsupervised GAN based inpainting framework.  ...  To our knowledge, this is the first demonstration of an unsupervised GAN based sequence inpainting.  ...  Improved GAN Samples and Reconstructions: Conditioning on structural priors forces the generator to yield samples closer to natural data manifold.  ... 
arXiv:1908.05861v1 fatcat:uppoevrpzzeshnya33dc2du4pi

Shape Inpainting using 3D Generative Adversarial Network and Recurrent Convolutional Networks [article]

Weiyue Wang, Qiangui Huang, Suya You, Chao Yang, Ulrich Neumann
2017 arXiv   pre-print
To inpaint 3D models with semantic plausibility and contextual details, we introduce a hybrid framework that combines a 3D Encoder-Decoder Generative Adversarial Network (3D-ED-GAN) and a Long-term Recurrent  ...  The 3D-ED-GAN is a 3D convolutional neural network trained with a generative adversarial paradigm to fill missing 3D data in low-resolution.  ...  Loss function The generator G in 3D-ED-GAN is modeled by the Encoder-Decoder network. This can be viewed as a conditional GAN, in which the latent distribution is conditioned on given context data.  ... 
arXiv:1711.06375v1 fatcat:gi6o36a2jrcfhamtdjqrhjec4y

Progressive Generative Adversarial Networks: Deep Learning in Head and Neck Cancer CT Images to Synthesized PET Images Generation for Hybrid PET/CT Application

Bin HUANG, Zhe-wei CHEN, Martin LAW, Shi-ting FENG, Qiao-liang LI, Bing-sheng HUANG
2018 DEStech Transactions on Computer Science and Engineering  
We proposed a progressive Generative Adversarial Networks (GAN) to generate synthesized positron emission tomography (PET) images from computed tomography (CT) images for hybrid PET/CT application.  ...  to noise ratio (PNSR) evaluation indicators, indicating that the proposed method is suitable for generating medical images to use in hybrid systems.  ...  U1713220), Shenzhen Municipal Scheme for Technology Research (No. JCYJ20170302152605463).  ... 
doi:10.12783/dtcse/ccnt2018/24701 fatcat:iao3h5acazdqpnfvvkxc7wxibm

A Hybrid GAN Based Approach to Solve Imbalanced Data Problem in Recommendation Systems

Wafa Shafqat, Yung-Cheol Byun
2022 IEEE Access  
We implemented conditional Wasserstein GAN with gradient penalty to generate tabular data containing both numerical and categorical values.  ...  In this paper, we propose a hybrid GAN approach to solve the data imbalance problem to enhance recommendation systems' performance.  ...  In our work, we take the concept of hybrid GAN architecture to generate synthetic click sequence data and oversample the minority class to enhance the performance of recommendation systems.  ... 
doi:10.1109/access.2022.3141776 fatcat:btxxp4vpy5gvxcbmzs2kc5jmlq

Combining Transformer Generators with Convolutional Discriminators [article]

Ricard Durall, Stanislav Frolov, Jörn Hees, Federico Raue, Franz-Josef Pfreundt, Andreas Dengel, Janis Keupe
2021 arXiv   pre-print
At the same time, image synthesis using generative adversarial networks (GANs) has drastically improved over the last few years.  ...  We evaluate our approach by conducting a benchmark of well-known CNN discriminators, ablate the size of the transformer-based generator, and show that combining both architectural elements into a hybrid  ...  Given the success of CNNs for vision problems, in this work we explore the combination of a purely transformer-based generator and CNN discriminator into a hybrid GAN for image synthesis.  ... 
arXiv:2105.10189v3 fatcat:wvise5ncufcvzmqgj5rdq2fi4q

Quantum Deep Learning for Mutant COVID-19 Strain Prediction [article]

Yu-Xin Jin, Jun-Jie Hu, Qi Li, Zhi-Cheng Luo, Fang-Yan Zhang, Hao Tang, Kun Qian, Xian-Min Jin
2022 arXiv   pre-print
The results state that the fidelities of random generating spike protein variation structure are always beyond 96% for Delta, 94% for Omicron.  ...  In addition, this hybrid quantum-classical model for the first time achieves quantum-inspired blur convolution similar to classical depthwise convolution and also successfully applies quantum progressive  ...  The algorithm flow of quantum style-based GAN We propose a hybrid quantum-classical model QuStyleGAN(quantum style-based GAN) [34, 38] for COVID-19 epidemic strain prediction, which directly takes spike  ... 
arXiv:2203.03556v1 fatcat:3wmo3inpkbf2lmvwoymvirwwhu

MuseGAN: Multi-track Sequential Generative Adversarial Networks for Symbolic Music Generation and Accompaniment

Hao-Wen Dong, Wen-Yi Hsiao, Li-Chia Yang, Yi-Hsuan Yang
2018 PROCEEDINGS OF THE THIRTIETH AAAI CONFERENCE ON ARTIFICIAL INTELLIGENCE AND THE TWENTY-EIGHTH INNOVATIVE APPLICATIONS OF ARTIFICIAL INTELLIGENCE CONFERENCE  
In this paper, we propose three models for symbolic multi-track music generation under the framework of generative adversarial networks (GANs).  ...  The three models, which differ in the underlying assumptions and accordingly the network architectures, are referred to as the jamming model, the composer model and the hybrid model.  ...  Conclusion In this work, we have presented a novel generative model for multi-track sequence generation under the framework of GANs.  ... 
doi:10.1609/aaai.v32i1.11312 fatcat:kwdhmswq7bghbjz4flxrld6hkm

Quantum Generative Adversarial Network: A Survey

Tong Li, Shi-bin Zhang, Jinyue Xia
2020 Computers Materials & Continua  
Generative adversarial network (GAN) is one of the most promising methods for unsupervised learning in recent years.  ...  The research involving QGAN The research of QGAN mainly involving: GAN, QML, quantum-classical hybrid model.  ...  Speech and text In the field of sequence data generation, GAN also has some applications.  ... 
doi:10.32604/cmc.2020.010551 fatcat:y6t752fx6be4fcxknmdavj5ed4

Chunked Autoregressive GAN for Conditional Waveform Synthesis [article]

Max Morrison, Rithesh Kumar, Kundan Kumar, Prem Seetharaman, Aaron Courville, Yoshua Bengio
2022 arXiv   pre-print
Generative adversarial networks (GANs) have become a common choice for non-autoregressive waveform synthesis.  ...  We show that simple pitch and periodicity conditioning is insufficient for reducing this error relative to using autoregression.  ...  Acknowledgments The authors would like to thank Jose Sotelo, Lucas Gestin, Vicki Anand, and Christian Schilter for valuable discussions and inputs.  ... 
arXiv:2110.10139v2 fatcat:rb3wjcja3jegrkedcb5ysmj75y

GAN Computers Generate Arts? A Survey on Visual Arts, Music, and Literary Text Generation using Generative Adversarial Network [article]

Sakib Shahriar
2021 arXiv   pre-print
This survey takes a comprehensive look at the recent works using GANs for generating visual arts, music, and literary text.  ...  Finally, some of the key challenges in art generation using GANs are highlighted along with recommendations for future work.  ...  [38] used conditional GANs for melody generation. The model consists of three layers.  ... 
arXiv:2108.03857v1 fatcat:6m6nhyv37zasxem2fpmcvv5nua

MuseGAN: Multi-track Sequential Generative Adversarial Networks for Symbolic Music Generation and Accompaniment [article]

Hao-Wen Dong, Wen-Yi Hsiao, Li-Chia Yang, Yi-Hsuan Yang
2017 arXiv   pre-print
In this paper, we propose three models for symbolic multi-track music generation under the framework of generative adversarial networks (GANs).  ...  The three models, which differ in the underlying assumptions and accordingly the network architectures, are referred to as the jamming model, the composer model and the hybrid model.  ...  Conclusion In this work, we have presented a novel generative model for multi-track sequence generation under the framework of GANs.  ... 
arXiv:1709.06298v2 fatcat:l5g3ey34lfdo3i6mag5dapoanq

HVTR: Hybrid Volumetric-Textural Rendering for Human Avatars [article]

Tao Hu, Tao Yu, Zerong Zheng, He Zhang, Yebin Liu, Matthias Zwicker
2022 arXiv   pre-print
The key advantage of our approach is that we can then convert the fused features into a high-resolution, high-quality avatar by a fast GAN-based textural renderer.  ...  We propose a novel neural rendering pipeline, Hybrid Volumetric-Textural Rendering (HVTR), which synthesizes virtual human avatars from arbitrary poses efficiently and at high quality.  ...  Rendering Humans by Generative Adversarial Network (GAN).  ... 
arXiv:2112.10203v2 fatcat:356of2nl7jc7thxo3bzcbrhhpm

Semi-Recurrent CNN-based VAE-GAN for Sequential Data Generation [article]

Mohammad Akbari, Jie Liang
2018 arXiv   pre-print
A semi-recurrent hybrid VAE-GAN model for generating sequential data is introduced.  ...  Two testing frameworks for synthesizing a sequence with any number of frames are also proposed.  ...  For example, Wasserstein GAN (WGAN) [18] used Wasserstein distance as an objective for training GANs to improve the stability of learning, Laplacian GAN (LAP-GANs) [19] achieved coarse-to-fine conditional  ... 
arXiv:1806.00509v1 fatcat:dtdqrmfml5e5pn4xgnx67qkpum

Data Augmentation Methods for End-to-end Speech Recognition on Distant-Talk Scenarios [article]

Emiru Tsunoo, Kentaro Shibata, Chaitanya Narisetty, Yosuke Kashiwagi, Shinji Watanabe
2021 arXiv   pre-print
We propose to use three augmentation methods and thier combinations: 1) data augmentation using text-to-speech (TTS) data, 2) cycle-consistent generative adversarial network (Cycle-GAN) augmentation trained  ...  to map two different audio characteristics, the one of clean speech and of noisy recordings, to match the testing condition, and 3) pseudo-label augmentation provided by the pretrained ASR module for  ...  Instead of shuffling speaker conditioning information, as in [22] , we perturb synthesized speech with various RIRs generated by a room simulator and additive noise for the CHiME-6 setup.  ... 
arXiv:2106.03419v1 fatcat:w4x6uf2k6rcchdr3rpowxjss7q
« Previous Showing results 1 — 15 out of 12,902 results