Escaping from Collapsing Modes in a Constrained Space

Chang, Chia-Che; Lin, Chieh Hubert; Lee, Che-Rung; Juan, Da-Cheng; Wei, Wei; Chen, Hwann-Tzong

doi:10.1007/978-3-030-01234-2_13

Cited by 16 publications

(12 citation statements)

References 12 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…This application shares similar context to some bijection methods [10,8,3], despite COCO-GAN estimates the latent vector with a single macro patch instead of the full image. In addition, the application is also similar to image restoration [17,33,34] or image out-painting [27].…”

Section: Patch-guided Image Generationmentioning

confidence: 76%

COCO-GAN: Generation by Parts via Conditional Coordinating

Lin

Chang

Chen

et al. 2019

2019 IEEE/CVF International Conference on Computer Vision (ICCV)

Self Cite

View full text Add to dashboard Cite

Humans can only interact with part of the surrounding environment due to biological restrictions. Therefore, we learn to reason the spatial relationships across a series of observations to piece together the surrounding environment. Inspired by such behavior and the fact that machines also have computational constraints, we propose COnditional COordinate GAN (COCO-GAN) of which the generator generates images by parts based on their spatial coordinates as the condition. On the other hand, the discriminator learns to justify realism across multiple assembled patches by global coherence, local appearance, and edge-crossing continuity. Despite the full images are never generated during training, we show that COCO-GAN can produce state-of-the-art-quality full images during inference. We further demonstrate a variety of novel applications enabled by teaching the network to be aware of coordinates. First, we perform extrapolation to the learned coordinate manifold and generate off-the-boundary patches. Combining with the originally generated full image, COCO-GAN can produce images that are larger than training samples, which we called "beyond-boundary generation". We then showcase panorama generation within a cylindrical coordinate system that inherently preserves horizontally cyclic topology. On the computation side, COCO-GAN has a built-in divide-and-conquer paradigm that reduces memory requisition during training and inference, provides highparallelism, and can generate parts of images on-demand.

show abstract

Section: Patch-guided Image Generationmentioning

confidence: 76%

COCO-GAN: Generation by Parts via Conditional Coordinating

Lin

Chang

Chen

et al. 2019

2019 IEEE/CVF International Conference on Computer Vision (ICCV)

Self Cite

View full text Add to dashboard Cite

show abstract

“…We conducted qualitative analysis by comparing the output of our model with those of two other auto-encoder-based GAN models, BEGAN [12] and BEGAN-CS [18]. BEGAN-CS adds a latent constraint to BEGAN.…”

Section: Qualitative Resultsmentioning

confidence: 99%

“…Section 2 presents the theoretical background and provides a detailed description of the proposed model. We then demonstrate the superiority of our model by qualitatively and quantitatively comparing it to conventional models [12,18] in Section 3. Concluding remarks are presented in Section 4.…”

Section: Introductionmentioning

confidence: 96%

Unsupervised Generation and Synthesis of Facial Images via an Auto-Encoder-Based Deep Generative Adversarial Network

Kwak

2020

Applied Sciences

View full text Add to dashboard Cite

The processing of facial images is an important task, because it is required for a large number of real-world applications. As deep-learning models evolve, they require a huge number of images for training. In reality, however, the number of images available is limited. Generative adversarial networks (GANs) have thus been utilized for database augmentation, but they suffer from unstable training, low visual quality, and a lack of diversity. In this paper, we propose an auto-encoder-based GAN with an enhanced network structure and training scheme for Database (DB) augmentation and image synthesis. Our generator and decoder are divided into two separate modules that each take input vectors for low-level and high-level features; these input vectors affect all layers within the generator and decoder. The effectiveness of the proposed method is demonstrated by comparing it with baseline methods. In addition, we introduce a new scheme that can combine two existing images without the need for extra networks based on the auto-encoder structure of the discriminator in our model. We add a novel double-constraint loss to make the encoded latent vectors equal to the input vectors.

show abstract

“…There are many studies on GANs [13][14][15]24,[32][33][34][35][36][37][38][39][40][41][42][43][44][45][46][47]. In this section, we investigate the studies that consider the stability problem during the training of GANs.…”

Section: Related Workmentioning

confidence: 99%

“…It was conducted by training a multi-layer neural network with sample data and the standard deviation of the data. Boundary Equilibrium Generative Adversarial Nets-Constrained Space (BEGAN-CS) added an embedding space-constrained loss in BEGAN and showed the stability improvement by using the proportional coefficient's variation during training [47]. Although the results of VEEGAN and BEGAN-CS were noticeable, they had limitations that require additional CNN models and a constraint of latent space for a discriminator, respectively.…”

Section: Stable Training and An Evaluation Indexmentioning

confidence: 99%

Simplified Fréchet Distance for Generative Adversarial Nets

Kim

Jung

et al. 2020

Sensors

View full text Add to dashboard Cite

We introduce a distance metric between two distributions and propose a Generative Adversarial Network (GAN) model: the Simplified Fréchet distance (SFD) and the Simplified Fréchet GAN (SFGAN). Although the data generated through GANs are similar to real data, GAN often undergoes unstable training due to its adversarial structure. A possible solution to this problem is considering Fréchet distance (FD). However, FD is unfeasible to realize due to its covariance term. SFD overcomes the complexity so that it enables us to realize in networks. The structure of SFGAN is based on the Boundary Equilibrium GAN (BEGAN) while using SFD in loss functions. Experiments are conducted with several datasets, including CelebA and CIFAR-10. The losses and generated samples of SFGAN and BEGAN are compared with several distance metrics. The evidence of mode collapse and/or mode drop does not occur until 3000k steps for SFGAN, while it occurs between 457k and 968k steps for BEGAN. Experimental results show that SFD makes GANs more stable than other distance metrics used in GANs, and SFD compensates for the weakness of models based on BEGAN-based network structure. Based on the experimental results, we can conclude that SFD is more suitable for GAN than other metrics.

show abstract

Escaping from Collapsing Modes in a Constrained Space

Cited by 16 publications

References 12 publications

COCO-GAN: Generation by Parts via Conditional Coordinating

COCO-GAN: Generation by Parts via Conditional Coordinating

Unsupervised Generation and Synthesis of Facial Images via an Auto-Encoder-Based Deep Generative Adversarial Network

Simplified Fréchet Distance for Generative Adversarial Nets

Contact Info

Product

Resources

About