进入正文
生成式对抗网络,即所谓的GAN是近些年来最火的无监督学习方法之一,模型由Goodfellow等人在2014年首次提出,将博弈论中非零和博弈思想与生成模型结合在一起,巧妙避开了传统生成模型中概率密度估计困难等问题,是生成模型达到良好的效果。本文总结收集了一些关于生成对抗网络的学习资源,有兴趣者可以好好学一学。
01
基础知识
台大李弘毅老师gan课程。参考链接:
#youtube#:
https://www.youtube.com/watch?v=DQNNMiAP5lw&index=1&list=PLJV_el3uVTsMq6JEFPW35BCiOQTsoqwNw
#bilibili#
https://www.bilibili.com/video/av24011528?from=search&seid=11459671583323410876
成对抗网络初学入门:一文读懂GAN的基本原理
http://www.xtecher.com/Xfeature/view?aid=7496
深入浅出:GAN原理与应用入门介绍
https://zhuanlan.zhihu.com/p/28731033
港理工在读博士李嫣然深入浅出GAN之应用篇https://pan.baidu.com/s/1o8n4UDk 密码: 78wt
萌物生成器:如何使用四种GAN制造猫图https://zhuanlan.zhihu.com/p/27769807
GAN学习指南:从原理入门到制作生成Demohttps://zhuanlan.zhihu.com/p/24767059x
生成式对抗网络GAN研究进展http://blog.csdn.net/solomon1558/article/details/52537114
02
相关报告
NIPS 2016教程:生成对抗网络
https://arxiv.org/pdf/1701.00160.pdf
训练GANs的技巧和窍门https://github.com/soumith/ganhacks
OpenAI生成模型https://blog.openai.com/generative-models/
03
论文前言
对抗实例的解释和利用(Explaining and Harnessing Adversarial Examples)2014https://arxiv.org/pdf/1412.6572.pdf
基于深度生成模型的半监督学习( Semi-Supervised Learning with Deep Generative Models )2014https://arxiv.org/pdf/1406.5298v2.pdf
条件生成对抗网络(Conditional Generative Adversarial Nets)2014https://arxiv.org/pdf/1411.1784v1.pdf
基于深度卷积生成对抗网络的无监督学习(Unsupervised Representation Learning with Deep Convolutional Generative Adversarial Networks (DCGANs))2015https://arxiv.org/pdf/1511.06434v2.pdf
基于拉普拉斯金字塔生成式对抗网络的深度图像生成模型(Deep Generative Image Models using a Laplacian Pyramid of Adversarial Networks)2015http://papers.nips.cc/paper/5773-deep-generative-image-models-using-a-5. laplacian-pyramid-of-adversarial-networks.pdf
生成式矩匹配网络(Generative Moment Matching Networks)2015http://proceedings.mlr.press/v37/li15.pdf
超越均方误差的深度多尺度视频预测(Deep multi-scale video prediction beyond mean square error)2015https://arxiv.org/pdf/1511.05440.pdf
通过学习相似性度量的超像素自编码(Autoencoding beyond pixels using a learned similarity metric)2015https://arxiv.org/pdf/1512.09300.pdf
对抗自编码(Adversarial Autoencoders)2015https://arxiv.org/pdf/1511.05644.pdf
基于像素卷积神经网络的条件生成图片(Conditional Image Generation with PixelCNN Decoders)2015https://arxiv.org/pdf/1606.05328.pdf
通过平均差异最大优化训练生成神经网络(Training generative neural networks via Maximum Mean Discrepancy optimization)2015https://arxiv.org/pdf/1505.03906.pdf
训练GANs的一些技巧(Improved Techniques for Training GANs)2016
https://arxiv.org/pdf/1606.03498v1.pdf
InfoGAN:基于信息最大化GANs的可解释表达学习(InfoGAN:Interpretable Representation Learning by Information Maximizing Generative Adversarial Nets)2016https://arxiv.org/pdf/1606.03657v1.pdf
上下文像素编码:通过修复进行特征学习(Context Encoders: Feature Learning by Inpainting)2016
http://www.cvfoundation.org/openaccess/content_cvpr_2016/papers/Pathak_Context_Encoders_Feature_CVPR_2016_paper.pdf
生成对抗网络实现文本合成图像(Generative Adversarial Text to Image Synthesis)2016http://proceedings.mlr.press/v48/reed16.pdf
对抗特征学习(Adversarial Feature Learning)2016https://arxiv.org/pdf/1605.09782.pdf
结合逆自回归流的变分推理(Improving Variational Inference with Inverse Autoregressive Flow )2016https://papers.nips.cc/paper/6581-improving-variational-autoencoders-with-inverse-autoregressive-flow.pdf
深度学习系统对抗样本黑盒攻击(Practical Black-Box Attacks against Deep Learning Systems using Adversarial Examples)2016https://arxiv.org/pdf/1602.02697.pdf
参加,推断,重复:基于生成模型的快速场景理解(Attend, infer, repeat: Fast scene understanding with generative models)2016https://arxiv.org/pdf/1603.08575.pdf
f-GAN: 使用变分散度最小化训练生成神经采样器(f-GAN: Training Generative Neural Samplers using Variational Divergence Minimization )2016http://papers.nips.cc/paper/6066-tagger-deep-unsupervised-perceptual-grouping.pdf
在自然图像流形上的生成视觉操作(Generative Visual Manipulation on the Natural Image Manifold)2016https://arxiv.org/pdf/1609.03552.pdf
对抗性推断学习(Adversarially Learned Inference)2016https://arxiv.org/pdf/1606.00704.pdf
基于循环对抗网络的图像生成(Generating images with recurrent adversarial networks)2016https://arxiv.org/pdf/1602.05110.pdf
生成对抗模仿学习(Generative Adversarial Imitation Learning)2016http://papers.nips.cc/paper/6391-generative-adversarial-imitation-learning.pdf
基于3D生成对抗模型学习物体形状的概率隐空间(Learning a Probabilistic Latent Space of Object Shapes via 3D Generative-Adversarial Modeling)2016https://arxiv.org/pdf/1610.07584.pdf
学习画画(Learning What and Where to Draw)2016https://arxiv.org/pdf/1610.02454v1.pdf
基于辅助分类器GANs的条件图像合成(Conditional Image Synthesis with Auxiliary Classifier GANs)2016https://arxiv.org/pdf/1610.09585.pdf
隐生成模型的学习(Learning in Implicit Generative Models)2016https://arxiv.org/pdf/1610.03483.pdf
VIME: 变分信息最大化探索(VIME: Variational Information Maximizing Exploration)2016http://papers.nips.cc/paper/6591-vime-variational-information-maximizing-exploration.pdf
生成对抗网络的展开(Unrolled Generative Adversarial Networks)2016https://arxiv.org/pdf/1611.02163.pdf
基于内省对抗网络的神经图像编辑(Neural Photo Editing with Introspective Adversarial Networks)2016,原文链接:
https://arxiv.org/pdf/1609.07093.pdf
基于解码器的生成模型的定量分析(On the Quantitative Analysis of Decoder-Based Generative Models )2016,原文链接:
https://arxiv.org/pdf/1611.04273.pdf
结合生成对抗网络和Actor-Critic 方法(Connecting Generative Adversarial Networks and Actor-Critic Methods)2016,原文链接:
https://arxiv.org/pdf/1610.01945.pdf
通过对抗网络使用模拟和非监督图像训练( Learning from Simulated and Unsupervised Images through Adversarial Training)2016,原文链接:
https://arxiv.org/pdf/1612.07828.pdf
基于上下文RNN-GANs的抽象推理图的生成(Contextual RNN-GANs for Abstract Reasoning Diagram Generation)2016,原文链接:
https://arxiv.org/pdf/1609.09444.pdf
生成多对抗网络(Generative Multi-Adversarial Networks)2016,原文链接:
https://arxiv.org/pdf/1611.01673.pdf
生成对抗网络组合(Ensembles of Generative Adversarial Network)2016,原文链接:
https://arxiv.org/pdf/1612.00991.pdf
改进生成器目标的GANs(Improved generator objectives for GANs) 2016,原文链接:
https://arxiv.org/pdf/1612.02780.pdf
训练生成对抗网络的基本方法(Towards Principled Methods for Training Generative Adversarial Networks)2017,原文链接:
https://arxiv.org/pdf/1701.04862.pdf
生成对抗模型的隐向量精准修复(Precise Recovery of Latent Vectors from Generative Adversarial Networks)2017,原文链接:
https://openreview.NET/pdf?id=HJC88BzFl
生成混合模型(Generative Mixture of Networks)2017,原文链接:
https://arxiv.org/pdf/1702.03307.pdf
记忆生成时空模型(Generative Temporal Models with Memory)2017,原文链接:
https://arxiv.org/pdf/1702.04649.pdf
AdaGAN: Boosting Generative Models". AdaGAN。原文链接[https://arxiv.org/abs/1701.04862]
Loss-Sensitive Generative Adversarial Networks on Lipschitz Densities。原文链接:https://arxiv.org/abs/1701.06264;代码:https://github.com/guojunq/glsgan/
Wasserstein GAN, WGAN。原文链接:[https://arxiv.org/abs/1701.07875];代码:
https://github.com/martinarjovsky/WassersteinGAN
Boundary-Seeking Generative Adversarial Networks,BSGAN。原文链接:https://arxiv.org/abs/1702.08431;代码地址:
https://github.com/wiseodd/generative-models
Generative Adversarial Nets with Labeled Data by Activation Maximization,AMGAN。原文链接:
https://arxiv.org/abs/1703.02000
Triple Generative Adversarial Nets,Triple-GAN。原文链接
https://arxiv.org/abs/1703.02291
BEGAN: Boundary Equilibrium Generative Adversarial Networks。原文链接:https://arxiv.org/abs/1703.10717;代码:
https://github.com/wiseodd/generative-models
Improved Training of Wasserstein GANs。原文链接:https://arxiv.org/abs/1704.00028;代码:
https://github.com/wiseodd/generative-models
MAGAN: Margin Adaptation for Generative Adversarial Networks。原文链接[https://arxiv.org/abs/1704.03817],
代码:https://github.com/wiseodd/generative-models
Gang of GANs: Generative Adversarial Networks with Maximum Margin Ranking。原文链接:
https://arxiv.org/abs/1704.04865
Softmax GAN。原文链接:
https://arxiv.org/abs/1704.06191
Generative Adversarial Trainer: Defense to Adversarial Perturbations with GAN。原文链接:
https://arxiv.org/abs/1705.03387
Flow-GAN: Bridging implicit and prescribed learning in generative models。原文链接:
https://arxiv.org/abs/1705.08868
Approximation and Convergence Properties of Generative Adversarial Learning。原文链接:
https://arxiv.org/abs/1705.08991
Towards Consistency of Adversarial Training for Generative Models。原文链接:
https://arxiv.org/abs/1705.09199
Good Semi-supervised Learning that Requires a Bad GAN。原文链接:https://arxiv.org/abs/1705.09783
On Unifying Deep Generative Models。原文链接:
https://arxiv.org/abs/1706.00550
DeLiGAN:Generative Adversarial Networks for Diverse and Limited Data。原文链接:
http://10.254.1.82/cache/6/03/openaccess.thecvf.com/e029768353404049dbcac9187a363d5a/Gurumurthy_DeLiGAN__Generative_CVPR_2017_paper.pdf;代码:https://github.com/val-iisc/deligan
Temporal Generative Adversarial Nets With Singular Value Clipping。原始链接:
http://openaccess.thecvf.com/content_ICCV_2017/papers/Saito_Temporal_Generative_Adversarial_ICCV_2017_paper.pdf
Least Squares Generative Adversarial Networks. LSGAN。原始链接:
http://openaccess.thecvf.com/content_ICCV_2017/papers/Mao_Least_Squares_Generative_ICCV_2017_paper.pdf
04
项目实战
深度卷积生成对抗模型(DCGAN)参考链接
https://github.com/Newmu/dcgan_code
用Keras实现MNIST生成对抗模型,参考链接:
https://oshearesearch.com/index.PHP/2016/07/01/mnist-generative-adversarial-model-in-keras/
用深度学习TensorFlow实现图像修复,参考链接:
http://bamos.github.io/2016/08/09/deep-completion/
TensorFlow实现深度卷积生成对抗模型(DCGAN),参考链接:
https://github.com/carpedm20/DCGAN-tensorflow
Torch实现深度卷积生成对抗模型(DCGAN),参考链接:
https://github.com/soumith/dcgan.torch
Keras实现深度卷积生成对抗模型(DCGAN),参考链接:
https://github.com/jacobgil/keras-dcgan
使用神经网络生成自然图像(Facebook的Eyescream项目),参考链接:
https://github.com/facebook/eyescream
对抗自编码(AdversarialAutoEncoder),参考链接:
https://github.com/musyoku/adversarial-autoencoder
利用ThoughtVectors 实现文本到图像的合成,参考链接:
https://github.com/paarthneekhara/text-to-image
对抗样本生成器(Adversarialexample generator),参考链接:
https://github.com/e-lab/torch-toolbox/tree/master/Adversarial
深度生成模型的半监督学习,参考链接:
https://github.com/dpkingma/nips14-ssl
GANs的训练方法,参考链接:
https://github.com/openai/improved-gan
生成式矩匹配网络(Generative Moment Matching Networks, GMMNs),参考链接:
https://github.com/yujiali/gmmn
对抗视频生成,参考链接:
https://github.com/dyelax/Adversarial_Video_Generation
基于条件对抗网络的图像到图像翻译(pix2pix)参考链接:
https://github.com/phillipi/pix2pix
对抗机器学习库Cleverhans, 参考链接:
https://github.com/openai/cleverhans
05
相关补充
生成对抗网络(GAN)专知荟萃。参考资料:
http://www.zhuanzhi.ai/topic/2001150162715950/awesome
The GAN Zoo千奇百怪的生成对抗网络,都在这里了。你没看错,里面已经有有近百个了。参考链接:
https://github.com/hindupuravinash/the-gan-zoo
gan资料集锦。参考链接:
https://github.com/nightrome/really-awesome-gan
gan在医学上的案例集锦:
https://github.com/xinario/awesome-gan-for-medical-imaging
gan应用集锦:
https://github.com/nashory/gans-awesome-applications
生成对抗网络(GAN)的前沿进展(论文、报告、框架和Github资源)汇总,参考链接:
http://blog.csdn.net/love666666shen/article/details/74953970
2018/11/15
Thursday
上面就是关于生成对抗网络-GAN的详细资料了,内容有点多,但绝对是干货满满,生成对抗网络作为无监督学习的典型,有着广阔的发展前景。后续还会继续分享关于机器学习、深度学习相关的知识,请记得继续关注哦!
往期回顾
●
●
●
●