Filters








92,863 Hits in 4.2 sec

On the Evaluation of Conditional GANs [article]

Terrance DeVries, Adriana Romero, Luis Pineda, Graham W. Taylor, Michal Drozdzal
2019 arXiv   pre-print
We conduct proof-of-concept experiments on a controllable synthetic dataset, which consistently highlight the benefits of FJD when compared to currently established metrics.  ...  Despite outstanding progress, quantitative evaluation of such models often involves multiple distinct metrics to assess different desirable properties, such as image quality, conditional consistency, and  ...  Moreover, the metric used to evaluate the forward model on the generated distribution depends on the conditioning modality and includes: accuracy in the case of class-conditioned generation, Intersection  ... 
arXiv:1907.08175v3 fatcat:4v6jqscdnzhwhh3t7xmtladmqi

Evaluating the impact of insect community on pine caterpillar density in different stand conditions

LI Tian-Sheng, WANG Guo-Hua, GAN Zhong-Nan, HUA Zheng-Yuan, ZHOU Guo-Fa, 1) Institute of Forest Protection, Chinese Academy of Forestry, Beijing 100091, 2) Department of Mathematics, Branch School of Peking University , Beijing 100083, 3) Forest Pest Control Station of Longyou county, Zhejiang Province, Longyou 324400, 4) Forest Pest Control Station of Quzhou City, Zhejiang Province, Quzhou 324002
1998 Biodiversity Science  
Evaluating the impact of insect community on pine caterpillar density in different stand conditions/ L I Tian2 Sheng 1) , ZHOU Guo2Fa 2) , WANG Guo2Hua 3) , GAN Zhong2Nan 4) , HUA Zheng2Yuan 4) Abstract  ...  The forest area was divided into four types according to t he variation of canopy and vegetation conditions by using cluster analysis , t here were significant differences in diversity index , species  ...  图中的一个点可能包含多个 样本点 Abscissa : canopy density ; Ordinate : vegetation conditions ; One point in t he figure may contain many samples 在第一步比较的基础上 ,下面比较 Ⅰ、 Ⅱ类及 Ⅲ、 Ⅳ类的差异 。表 3 是比较的结果 。  ... 
doi:10.17520/biods.1998024 fatcat:4ocvuoyaivbzdaregblrzmhn5i

On Conditioning GANs to Hierarchical Ontologies [article]

Hamid Eghbal-zadeh, Lukas Fischer, Thomas Hoch
2019 arXiv   pre-print
Additionally, we show that the O-GAN achieves better conditioning results evaluated by implicit similarity between the text and the generated image.  ...  To handle the complexities of fashion image and meta data, we propose Ontology Generative Adversarial Networks (O-GANs) for fashion image synthesis that is conditioned on an hierarchical fashion ontology  ...  To evaluate the quality of the conditioning, we report the cross-entropy between the conditioning labels and the probability of the labels for generated images, estimated via the label predictor L.  ... 
arXiv:1905.06586v1 fatcat:s5646pmm3bftbgqbfiw3vos37i

Evaluation of the Initial Stage of Formation of Ti/Al Ohmic Contacts Using Photoresponse Method

Kenji Shiojima, Hideo Yokohama, Gako Araki
2013 Japanese Journal of Applied Physics  
Ti/Al contacts formed on n-GaN and AlGaN/GaN layers upon annealing at temperatures below the melting point of Al were evaluated by photoresponse (PR), current-voltage (I-V), and secondary ion mass spectroscopy  ...  In contrast, the n-GaN samples had very low q B values of 0.2 eV under the as-deposited condition.  ...  Acknowledgment Part of this work was supported by a Grant-in-Aid for Scientific Research (C) from the Ministry of Education, Culture, Sports, Science and Technology.  ... 
doi:10.7567/jjap.52.08jn06 fatcat:dgqz4exjxjhvvlojswckiah5cu

On Enhancing Speech Emotion Recognition Using Generative Adversarial Networks

Saurabh Sahu, Rahul Gupta, Carol Espy-Wilson
2018 Interspeech 2018  
that learns the distribution of the higher dimensional feature vectors conditioned on the labels or the emotional class to which it belongs.  ...  Specifically, we investigate two set ups: (i) a vanilla GAN that learns the distribution of a lower dimensional representation of the actual higher dimensional feature vector and, (ii) a conditional GAN  ...  Conditional GAN is one such example where the synthetic data generation is conditioned on labels.  ... 
doi:10.21437/interspeech.2018-1883 dblp:conf/interspeech/SahuGE18 fatcat:lt3wbdycpjaetm47vflr7vhtgq

On Enhancing Speech Emotion Recognition using Generative Adversarial Networks [article]

Saurabh Sahu, Rahul Gupta, Carol Espy-Wilson
2018 arXiv   pre-print
that learns the distribution of the higher dimensional feature vectors conditioned on the labels or the emotional class to which it belongs.  ...  Specifically, we investigate two set ups: (i) a vanilla GAN that learns the distribution of a lower dimensional representation of the actual higher dimensional feature vector and, (ii) a conditional GAN  ...  Conditional GAN is one such example where the synthetic data generation is conditioned on labels.  ... 
arXiv:1806.06626v1 fatcat:l254bomkqncl3bylf3frag2wdi

Deep learning approach to generate a synthetic cognitive psychology behavioral dataset

Jung-gu Choi, Yoonjin Nah, Inhwan Ko, Sanghoon Han
2021 IEEE Access  
In the case of the overlapped sample test at the instance level evaluation, we evaluated different samples in the generated dataset based on one-sample t-test results.  ...  The difference values of the GAN-based model conditions were generally lower than those of the random generation condition.  ...  YOONJIN NAH is currently a postgraduate researcher in the Department of Psychology at Yonsei University.  ... 
doi:10.1109/access.2021.3120083 fatcat:gtbrld3rmfbypi4zsv5xunaicy

Semantic Bottleneck Scene Generation [article]

Samaneh Azadi, Michael Tschannen, Eric Tzeng, Sylvain Gelly, Trevor Darrell, Mario Lucic
2019 arXiv   pre-print
For the latter, we use a conditional segmentation-to-image synthesis network that captures the distribution of photo-realistic images conditioned on the semantic layout.  ...  Coupling the high-fidelity generation capabilities of label-conditional image synthesis methods with the flexibility of unconditional generative models, we propose a semantic bottleneck GAN model for unconditional  ...  We thank Marvin Ritter for help with issues related to the compare gan library [27] . We are grateful to the members of BAIR for fruitful discussions.  ... 
arXiv:1911.11357v1 fatcat:id7o6lwt6bejfcs2tlnxt2rrb4

StudioGAN: A Taxonomy and Benchmark of GANs for Image Synthesis [article]

Minguk Kang, Joonghyuk Shin, Jaesik Park
2022 arXiv   pre-print
Generative Adversarial Network (GAN) is one of the state-of-the-art generative models for realistic image synthesis.  ...  We study the taxonomy of GAN approaches and present a new open-source library named StudioGAN.  ...  Evaluation metrics Currently, GAN evaluation hugely depends on the value of Fréchet Inception Distance (FID) [28] .  ... 
arXiv:2206.09479v2 fatcat:jlf4rshpwrfiflthqdkio226qu

Towards Diverse and Natural Image Descriptions via a Conditional GAN [article]

Bo Dai, Sanja Fidler, Raquel Urtasun, Dahua Lin
2017 arXiv   pre-print
Specifically, we propose a new framework based on Conditional Generative Adversarial Networks (CGAN), which jointly learns a generator to produce descriptions conditioned on images and an evaluator to  ...  In this paper, we explore an alternative approach, with the aim to improve the naturalness and diversity -- two essential properties of human expression.  ...  Towards this goal, we develop a new framework on top of the Conditional GAN [22] . GAN has been successfully used in image generation.  ... 
arXiv:1703.06029v3 fatcat:wdsgvpttrzdn5asxgg6xzxtbwq

Using generative adversarial networks to evaluate robustness of reinforcement learning agents against uncertainties

Fazel Khayatian, Zoltán Nagy, Andrew Bollinger
2021 Energy and Buildings  
The synthetic profiles are utilized as a resource for evaluating the response of trained machine learning models to unseen events.  ...  Highlights  GAN is utilized to create synthetic building performance profiles  Synthetic projections are conditioned based on climate and building operation  Uncertainty is infused into synthetic data  ...  Acknowledgement This research has been carried out within the context of the Empa-funded project Algorithmic Regulation and Control.  ... 
doi:10.1016/j.enbuild.2021.111334 fatcat:imbffqbkardobfs52pg7czzuym

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
A performance comparison and description of the various GAN architecture are also presented.  ...  Finally, some of the key challenges in art generation using GANs are highlighted along with recommendations for future work.  ...  GANs on the other hand are not restricted to any such conditions and therefore has the potential to create more realistic arts.  ... 
arXiv:2108.03857v1 fatcat:6m6nhyv37zasxem2fpmcvv5nua

Chunked Autoregressive GAN for Conditional Waveform Synthesis [article]

Max Morrison, Rithesh Kumar, Kundan Kumar, Prem Seetharaman, Aaron Courville, Yoshua Bengio
2022 arXiv   pre-print
However, state-of-the-art GAN-based models produce artifacts when performing mel-spectrogram inversion.  ...  Conditional waveform synthesis models learn a distribution of audio waveforms given conditioning such as text, mel-spectrograms, or MIDI.  ...  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

Conditioning of three-dimensional generative adversarial networks for pore and reservoir-scale models [article]

Lukas Mosser, Olivier Dubrule, Martin J. Blunt
2018 arXiv   pre-print
Based on the previous work of Yeh et al. (2016), we use a content loss to constrain to the conditioning data and a perceptual loss obtained from the evaluation of the GAN discriminator network.  ...  This contribution leverages the differentiable nature of neural networks to extend GANs to the conditional simulation of three-dimensional pore- and reservoir-scale models.  ...  Menke for providing the Ketton micro-CT image dataset as well as G. Mariethoz and J. Caers for sharing the Maules Creek training image. O.  ... 
arXiv:1802.05622v1 fatcat:we5llg6hjbft3pgakufr2ust3e

CTAB-GAN: Effective Table Data Synthesizing [article]

Zilong Zhao, Aditya Kunar, Hiek Van der Scheer, Robert Birke, Lydia Y. Chen
2021 arXiv   pre-print
We extensively evaluate CTAB-GAN with the state of the art GANs that generate synthetic tables, in terms of data similarity and analysis utility.  ...  The results on five datasets show that the synthetic data of CTAB-GAN remarkably resembles the real data for all three types of variables and results into higher accuracy for five machine learning algorithms  ...  Since our algorithm is based on conditional GAN, the generator requires a noise vector plus a conditional vector. Details on the conditional vector are given in Sec. 3.4.  ... 
arXiv:2102.08369v2 fatcat:kkfq3x7l3fabtnxcbc265snn3u
« Previous Showing results 1 — 15 out of 92,863 results