David Gan - Internet Archive Scholar

In this work, we present a method, GAN Sketching, for rewriting GANs with one or more sketches, to make GANs training easier for novice users. ... Finally, we demonstrate a few applications of the resulting GAN, including latent space interpolation and image editing. ... This choice has also been shown effective in previous few-shot GAN works [65] . ...

arXiv:2108.02774v2 fatcat:cuugqdqhr5catg4uqmf7dv4ehq

Open Access Multiple Versions

In this paper, we present three amorphous GaN models obtained from the first principles simulation. We find that a chemically ordered continuous random network is the ideal structure for a-GaN. ... In 1997, we proposed that a-GaN might find use as an electronic material 10, 11 . The structure of a-GaN is controversial. ... The VODS of crystal wurtzite GaN is plotted as an inset from Ref. 17. ...

arXiv:1103.1634v1 fatcat:pxded53fbrectoclcmqvn3lyze

GAN. ... We show that this interpretation more generally holds for f-GAN and Wasserstein GAN. As a byproduct, we apply the duality framework to a hybrid of f-divergence and Wasserstein distance. ... [38] Meisam Razaviyayn, Farzan Farnia, and David Tse. Discrete rényi classifiers. In Advances in Neural Information Processing Systems, pages 3276-3284, 2015. [39] Farzan Farnia and David Tse. ...

arXiv:1810.11740v1 fatcat:uqurcy24lbfgthquds6fyjm24u

Even in this simple benchmark, the GAN problem has not been well-understood as we observe that existing state-of-the-art GAN architectures may fail to learn a proper generative distribution owing to (1 ... Generative Adversarial Networks (GANs) have become a popular method to learn a probability model from data. ... +Spectral Normalization [14] , WGAN+Truncated Gradient Penalty [15] , relaxed WGAN [16] , f -GAN [17] , MMD-GAN [18, 19] , Least-Squares GAN [20] , Boundary equilibrium GAN [21] , etc.), and (3 ...

arXiv:1710.10793v2 fatcat:j22dyunjive55l3iyeazi35nia

Multiple Versions

Even then, some success was achieved using an established GAN approach. ... While the GAN managed to restore a credible face in some cases, this rarely resembled the original face. ...

arXiv:1810.06455v2 fatcat:szdwls56mvbb7jka4yipbnjfv4

Open Access Multiple Versions

Even then, some success was achieved using an established GAN approach. ... While the GAN managed to restore a credible face in many cases, this rarely resembled the original face. ...

doi:10.1101/447102 fatcat:au5czvn7hnfhvd7ipvtqi43fha

Many GAN image detectors have been proposed, recently. In real world scenarios, however, most of them show limited robustness and generalization ability. ... We investigate these problems and propose a new GAN image detector based on a limited sub-sampling architecture and a suitable contrastive learning paradigm. ... However, the decision threshold itself is unknown and, more important, it varies largely from GAN to GAN. ...

arXiv:2112.12606v1 fatcat:fe7ywost3faw7jpd2t67mxbkte

Nine microns of unintentionally doped (UID) GaN was then grown on the GaN substrates for all samples. ... GaN devices grown on GaN substrates have greatly reduced dislocation density (<10 6 cm -2 ) compared to heteroepitaxial GaN (10 8 -10 10 cm -2 ), which should assist in improving device performance for ... Figure 2 . 2 Cross-sectional TEM images of GaN-on-GaN Schottky diodes: (a) Sample NE showing no visible sign of GaN-GaN interface; (b) Sample E25 showing faint signs of small precipitates in p-GaN layer ...

doi:10.1017/s1431927619011930 fatcat:xgisixrexnep3caszwxg2n33xm

Faces generated using generative adversarial networks (GANs) have reached unprecedented realism. ... For security and criminal identification, the ability to provide a GAN-based system that works like a sketch artist would be incredibly useful. ... Which improvisations to this GAN model research have generated realistic images [11, 12] . Image-to-image translation is also an application of GANs. ...

arXiv:2107.10756v2 fatcat:bidckdrvafhhhpgdcvpsaidcea

Open Access Multiple Versions

AlN/GaN heterostructures with 1700-cm 2 /V · s Hall mobility have been grown by molecular beam epitaxy on freestanding GaN substrates. ... GaN substrates provide a way of achieving high mobility, high v e , and high RF performance in AlN/GaN transistors. ... CONCLUSION To the best of our knowledge, this is the first demonstration of an AlN/GaN MOS-HEMT on freestanding GaN. ...

doi:10.1109/led.2012.2228463 fatcat:jckjdxbfsrhr5h2ixvre4rqnsm

Our work opens the door for better understanding of GAN and the ability to select the best samples from a GAN model. ... Despite an impressive performance from the latest GAN for generating hyper-realistic images, GAN discriminators have difficulty evaluating the quality of an individual generated sample. ... Error Detection for GAN To discover errors generated by GAN, we must learn the difference between real distribution r and generated distribution g akin to a GAN discriminator. ...

arXiv:1912.00527v1 fatcat:qyzf7ix3mbfklackr2xrezpogu

Hybrid green light-emitting diodes (LEDs) comprised of n-ZnO/(InGaN/GaN) multi-quantum-wells/p-GaN were grown on semi-insulating AlN/sapphire using pulsed laser deposition for the n-ZnO and metal organic ... Each period consisted of a 2nm-thick InGaN quantum well with a 4-nm-thick GaN barrier. Finally, a n-ZnO layer 15 was grown on top of the (InGaN/GaN) MQW/p-GaN by PLD to complete the green LED. ... In this work, we adopted an inverted LED structure employing a n-ZnO layer grown on top of (InGaN/GaN) MQW/p-GaN/AlN/sapphire. ...

doi:10.1117/12.817033 fatcat:4pmnmpfvmjdtlcaj74pbolmvh4

GANs have also been used for medical image segmentation. ... 3D Pix2Pix GAN, named Vox2Vox, to segment brain gliomas. ...

arXiv:2003.13653v3 fatcat:l7r32mkujvc2voxct5fxeypfji

Open Access Multiple Versions

We therefore propose a new story-to-image-sequence generation model, StoryGAN, based on the sequential conditional GAN framework. ... In this task, Story-GAN significantly outperforms the other baselines. ... To ensure consistency across the sequence of the generated images, we adopt a two-level GAN framework. ...

arXiv:1812.02784v2 fatcat:jl3vtrrm55afhag5cnzrfinve4

Multiple Versions

We train a class-conditional GAN model without using manually annotated class labels. ... GAN-Tree uses clustering to split a GAN into a tree hierarchy of GANs for better mode coverage [29] . ... We suspect that this is because the GAN regularization loss stabilizes GAN training, preventing cluster instability from hurting GAN training stability. ...

arXiv:2006.10728v2 fatcat:xxgqfhnrobck3ecd73dinsdcky

Open Access Multiple Versions

Sketch Your Own GAN [article]

Preserved Fulltext

The properties of amorphous GaN [article]

Preserved Fulltext

A Convex Duality Framework for GANs [article]

Preserved Fulltext

Understanding GANs: the LQG Setting [article]

Preserved Fulltext

Other Versions

Refacing: reconstructing anonymized facial features using GANs [article]

Preserved Fulltext

Other Versions

Refacing: Reconstructing Anonymized Facial Features Using Gans [article]

Preserved Fulltext

Towards Universal GAN Image Detection [article]

Preserved Fulltext

Characterization of Etched and Grown GaN-GaN Schottky Diodes

Preserved Fulltext

Semantic Text-to-Face GAN -ST^2FG [article]

Preserved Fulltext

Other Versions

High Electron Velocity Submicrometer AlN/GaN MOS-HEMTs on Freestanding GaN Substrates

Preserved Fulltext

Detecting GAN generated errors [article]

Preserved Fulltext

Hybrid green LEDs based on n -ZnO/(InGaN/GaN) multi-quantum-wells/ p -GaN

Preserved Fulltext

Vox2Vox: 3D-GAN for Brain Tumour Segmentation [article]

Preserved Fulltext

Other Versions

StoryGAN: A Sequential Conditional GAN for Story Visualization [article]

Preserved Fulltext

Diverse Image Generation via Self-Conditioned GANs [article]

Preserved Fulltext