6 minute read


图1:GAN 的 论文数量 

1 学术

第一篇论文

  1. :o: Generative Adversarial Nets
    2014-06-10 Paper | Code

1.1 综述

  1. :o: Goodfellow I. NIPS 2016 tutorial: generative adversarial networks. arXiv preprint arXiv: 1701.00160, 2016.
    2016-12-31

  2. How Generative Adversarial Networks and its variants Work: An Overview of GAN
    2017-11 Paper, 中文笔记

  3. 生成式对抗网络 GAN 的研究进展与展望
    2017 中科院自动化所,中文综述;

1.2 理论

  1. Energy-based generative adversarial network
    Code
    Lecun paper

  2. Improved Techniques for Training GANs
    Code
    Goodfellow’s paper

  3. Mode Regularized Generative Adversarial Networks
    Yoshua Bengio , ICLR 2017

  4. Improving Generative Adversarial Networks with Denoising Feature Matching
    Code
    Yoshua Bengio , ICLR 2017

  5. Sampling Generative Networks
    Code

  6. How to train Gans

  7. Towards Principled Methods for Training Generative Adversarial Networks
    ICLR 2017

  8. Unrolled Generative Adversarial Networks
    Code
    ICLR 2017

  9. Least Squares Generative Adversarial Networks
    Code
    ICCV 2017

  10. Wasserstein GAN
    Code

  11. Improved Training of Wasserstein GANs
    Code
    The improve of wgan

  12. Towards Principled Methods for Training Generative Adversarial Networks

  13. Generalization and Equilibrium in Generative Adversarial Nets
    ICML 2017

  14. GANs Trained by a Two Time-Scale Uplast_modified_at Rule Converge to a Local Nash Equilibrium
    Code

  15. Spectral Normalization for Generative Adversarial Networks
    Code
    ICLR 2018

  16. Which Training Methods for GANs do actually Converge
    Code
    ICML 2018

  17. Self-Supervised Generative Adversarial Networks
    Code
    CVPR 2019

  18. Generative Flow via Invertible nxn Convolution
    2019-05-24 paper

1.3 经典网络

  1. StarGAN: Unified Generative Adversarial Networks for Multi-Domain Image-to-Image Translation
    2017-11-24 paper | pytorch
    多领域图像转换
  2. Unsupervised Representation Learning with Deep Convolutional Generative Adversarial Networks
    ICLR 2015 2015-11-19 Theano | Keras | Pytorch | Pytorch-MNIST/CelebA | Tensorflow | Torch
    DCGAN
    将卷积网络引入 GAN 中,且使用了 BN,证明了池化在 GAN 中不能使用;提供了许多有趣的生成结果;

1.4 模型评估

  1. An empirical study on evaluation metrics of generative adversarial networks
    2018-06 paper

  2. Pros and cons of gan evaluation measures
    2018-02 paper

  3. How good is my GAN?
    ECCV 2018 2018-07 paper

1.5 不确定性

GAN 生成图像的不确定性——不连贯;

  1. Visual Indeterminacy in Generative Neural Art
    NIPS 2019 workshop 2019-10-10 paper

2 技术应用

2.1 半监督学习

  1. Adversarial Training Methods for Semi-Supervised Text Classification
    Note
    Ian Goodfellow Paper

  2. Improved Techniques for Training GANs
    Code
    Goodfellow’s paper

  3. Unsupervised and Semi-supervised Learning with Categorical Generative Adversarial Networks
    ICLR

  4. Semi-Supervised QA with Generative Domain-Adaptive Nets
    ACL 2017

  5. Good Semi-supervised Learning that Requires a Bad GAN
    Code
    NIPS 2017

2.2 集成学习

  1. AdaGAN: Boosting Generative Models
    [Code] Google Brain

  2. GP-GAN: Towards Realistic High-Resolution Image Blending
    Code

2.3 条件对抗

  1. Conditional Generative Adversarial Nets
    Code

  2. InfoGAN: Interpretable Representation Learning by Information Maximizing Generative Adversarial Nets
    Code Code

  3. Conditional Image Synthesis With Auxiliary Classifier GANs
    Code
    GoogleBrain ICLR 2017

  4. Pixel-Level Domain Transfer
    Code

  5. Invertible Conditional GANs for image editing
    Code

  6. Plug & Play Generative Networks: Conditional Iterative Generation of Images in Latent Space
    Code

  7. Lifelong GAN: Continual Learning for Conditional Image Generation
    ICCV 2019 2019-07-23 paper

2.4 强化学习

  1. Connecting Generative Adversarial Networks and Actor-Critic Methods
    NIPS 2016 workshop

2.5 RNN

  1. C-RNN-GAN: Continuous recurrent neural networks with adversarial training
    Code

  2. SeqGAN: Sequence Generative Adversarial Nets with Policy Gradient
    Code(AAAI 2017)

2.6 离散分布生成

  1. Maximum-Likelihood Augmented Discrete Generative Adversarial Networks

  2. Boundary-Seeking Generative Adversarial Networks

  3. GANS for Sequences of Discrete Elements with the Gumbel-softmax Distribution

3 业务

3.1 CV

3.1.1 高质量生成

3.1.1.1 通用图像生成资源
3.1.1.2 人脸生成
3.1.1.3 人脸编辑
3.1.1.4 季节变换
3.1.1.5 风格迁移资源
3.1.1.6 其他
  1. Generative Adversarial Text to Image Synthesis
    Code Code

  2. Improved Techniques for Training GANs
    Code
    Goodfellow’s paper

  3. Plug & Play Generative Networks: Conditional Iterative Generation of Images in Latent Space
    Code

  4. StackGAN: Text to Photo-realistic Image Synthesis with Stacked Generative Adversarial Networks
    2016-11-10 paper | tensorflow | pytorch | v2-pytorch
    StackGAN:

  5. Improved Training of Wasserstein GANs
    Code

  6. Boundary Equibilibrium Generative Adversarial Networks Implementation in Tensorflow
    Code

  7. Progressive Growing of GANs for Improved Quality, Stability, and Variation
    Code[Tensorflow Code]

  8. Self-Attention Generative Adversarial Networks
    Code
    NIPS 2018

  9. Large Scale GAN Training for High Fidelity Natural Image Synthesis
    ICLR 2019

  10. A Style-Based Generator Architecture for Generative Adversarial Networks
    Code

  11. Dilated Spatial Generative Adversarial Networks for Ergodic Image Generation
    2019-05-15 paper
    膨胀卷积助力边缘清晰;

  12. Variational Hetero-Encoder Randomized Generative Adversarial Networks for Joint Image-Text Modeling
    2019-05-18 paper | OpenReview

3.1.2 检测

  1. Perceptual generative adversarial networks for small object detection
    CVPR 2017

  2. A-Fast-RCNN: Hard Positive Generation via Adversary for Object Detection
    Code
    CVPR 2017

3.1.3 分类&识别

  1. Generative OpenMax for Multi-Class Open Set Classification
    BMVC 2017

  2. Controllable Invariance through Adversarial Feature Learning
    Code
    NIPS 2017

  3. Unlabeled Samples Generated by GAN Improve the Person Re-identification Baseline in vitro
    Code
    ICCV2017

  4. Learning from Simulated and Unsupervised Images through Adversarial Training
    Code(Apple paper, CVPR 2017 Best Paper)

3.1.4 数据增强

  1. LiDAR Sensor modeling and Data augmentation with GANs for Autonomous driving
    ICML Workshop on AI for Autonomous Driving 2019 2019-05-17 paper

3.1.5 显著性检测

Saliency Prediction

  1. SalGAN: Visual Saliency Prediction with Generative Adversarial Networks
    2017-01-04 paper | theano

3.1.6 检索

  1. IRGAN: A Minimax Game for Unifying Generative and Discriminative Information Retrieval Models
    2017-05-30 paper
    IRGAN:

3.1.7 异常检测

Anomaly Detection

  1. Unsupervised Anomaly Detection with Generative Adversarial Networks to Guide Marker Discovery
    IPMI 2017 2017-03-17 paper
    医疗;

3.2 视频

3.2.1 视频生成

  1. Deep multi-scale video prediction beyond mean square error
    Code
    Yann LeCun’s paper

  2. Generating Videos with Scene Dynamics
    Web Code

  3. MoCoGAN: Decomposing Motion and Content for Video Generation

3.3 NLP

3.3.1 图像翻译

  1. UNSUPERVISED CROSS-DOMAIN IMAGE GENERATION
    Code

  2. Image-to-image translation using conditional adversarial nets
    Code
    Code

  3. Learning to Discover Cross-Domain Relations with Generative Adversarial Networks
    Code
  4. Unpaired Image-to-Image Translation using Cycle-Consistent Adversarial Networks
    ICCV 2017 2017-03-30 paper | torch-official | pytorch-official | project | blog
    CycleGAN

  5. CoGAN: Coupled Generative Adversarial Networks
    Code
    NIPS 2016

  6. Unsupervised Image-to-Image Translation with Generative Adversarial Networks
    NIPS 2017

  7. Unsupervised Image-to-Image Translation Networks

  8. Triangle Generative Adversarial Networks

  9. High-Resolution Image Synthesis and Semantic Manipulation with Conditional GANs
    Code
    掩码转图像;

  10. XGAN: Unsupervised Image-to-Image Translation for Many-to-Many Mappings
    Reviewed

  11. UNIT: UNsupervised Image-to-image Translation Networks
    Code
    NIPS 2017

  12. Toward Multimodal Image-to-Image Translation
    Code NIPS 2017

  13. Multimodal Unsupervised Image-to-Image Translation
    Code

  14. Video-to-Video Synthesis
    Code

  15. Everybody Dance Now
    Code

  16. GestureGAN for Hand Gesture-to-Gesture Translation in the Wild
    Code

  17. Art2Real: Unfolding the Reality of Artworks via Semantically-Aware Image-to-Image Translation
    CVPR 2019

  18. Fonts-2-Handwriting: A Seed-Augment-Train framework for universal digit classification
    ICLR 2019 2019-05-16 paper | code
    手写数字生成;

  19. Toward Learning a Unified Many-to-Many Mapping for Diverse Image Translation
    2019-05-21 paper

3.4 音频

3.4.1 MUSIC

  1. MidiNet: A Convolutional Generative Adversarial Network for Symbolic-domain Music Generation using 1D and 2D Conditions
    HOMEPAGE

3.5 多模态

3.5.1 融合

  1. M3D-GAN: Multi-Modal Multi-Domain Translation with Universal Attention
    2019-07-09 paper

3.5.2 语义图转照片

  1. S-Flow GAN
    2019-05-21 paper

4 其他

  1. Deep Generative Image Models using a Laplacian Pyramid of Adversarial Networks
    2015-01-18 paper | torch

  2. Adversarial Autoencoders
    2015-11-18 paper | chainer

  3. Generating Images with Perceptual Similarity Metrics based on Deep Networks
    2016-02-08 paper | tensorflow
    DeepSim

  4. Generating images with recurrent adversarial networks
    2016-02-16 paper | theano
    GRAN

  5. Generative Visual Manipulation on the Natural Image Manifold
    ECCV 2016 2016-09-12 paper | theano
    iGAN

  6. Learning What and Where to Draw
    NIPS 2016 2016-10-08 paper | torch

  7. Adversarial Training for Sketch Retrieval
    ECCV 2016 2016-07-10 paper

  8. Generative Image Modeling using Style and Structure Adversarial Networks
    2016-03-17 paper | torch

  9. Synthesizing the preferred inputs for neurons in neural networks via deep generator networks
    NIPS 2016-05-30 paper | caffe | project
    synthesizing

  10. Adversarial Feature Learning
    ICLR 2017 2016-05-31 paper

  11. Adversarially Learned Inference
    2016-06-02 paper | theano
    ALI

  12. Generative Adversarial Networks as Variational Training of Energy Based Models
    ICLR 2017 2016-11-06 paper | theano-offical


TOP

附录

A 数据集

B 研究员

  1. Ian Goodfellow
  2. GAN 汇总
  3. Awesome Adversarial Machine Learning
  4. TaehoonKim github
  5. Kuntal Ganguly
  6. Adversarial network(inFERENCe)
  7. InfoGan
  8. Deconvolution and Image Generation
  9. Gan theory(yingzhenli)
  10. Generative model(OpenAI)

C 报告

  1. Ian Goodfellow 演讲
  2. OpenAI

D 参考资料

a 论文

  1. GAN Zoo 汇总了所有的 GANs;

  2. AdversarialNetsPapers
    GANs 论文分类汇总;

  3. GAN Timeline
    GANs 项目汇总;

  4. GAN 论文汇总(韩东)

b 代码

  1. GANotebooks

  2. generative-models
    pytorch 和 tensorflow 实现的 GAN 和 VAE;

c 技能

  1. GAN 训练技巧 How to Train a GAN?

d 书籍

  1. 《GANs in Action》
    akub Langr, Vladimir Bok. GANs in Action[M]. -. 2019
    主页

  2. 《百面机器学习》
    诸葛越. 百面机器学习[M]. 北京:人民邮电出版社. 2018.298-332

  3. 《Learning GAN》·《GAN 实战生成对抗网络》
    Kuntal G. 著, 刘梦馨 译. GAN 实战生成对抗网络[M]. 北京:电子工业出版社, 2018.
    英文版本

  4. 《生成对抗网络入门指南》
    史丹青. 生成对抗网络入门指南[M]. 北京:机械工业出版社, 2018.

  5. 《Web 安全之强化学习与 GAN》
    刘淼. Web 安全之强化学习与 GAN[M]. 北京:机械工业出版社, 2018.

e 课程

[1]. 李宏毅. 李宏毅对抗生成网络(GAN)(2018)[EB/OL]. https://www.bilibili.com/video/av24011528?from=search&seid=14571953395351333549.
课件

Comments