不可錯過的 GAN 資源:教程、視頻、代碼實現、89 篇論文下載
新智元編譯
目錄
Workshops
教程 & 博客
論文
理論 & 機器學習
視覺應用
其他應用
幽默
視頻
代碼
Workshops
NIP 2016 對抗訓練 Workshop
【網頁】https://sites.google.com/site/nips2016adversarial/
【博客】http://www.inference.vc/my-summary-of-adversarial-training-nips-workshop/
教程 & 博客
如何訓練 GAN? 讓 GAN 工作的提示和技巧
【博客】https://github.com/soumith/ganhacks
NIPS 2016 教程:生成對抗網路
【arXiv】https://arxiv.org/abs/1701.00160
深度學習和 GAN 背後的直覺知識——一個基礎理解
【博客】https://blog.waya.ai/introduction-to-gans-a-boxing-match-b-w-neural-nets-b4e5319cc935
OpenAI——生成模型
【博客】https://openai.com/blog/generative-models/
SimGANs——無監督學習的遊戲規則顛覆者,無人車等
【博客】https://blog.waya.ai/simgans-applied-to-autonomous-driving-5a8c6676e36b
論文
理論 & 機器學習
生成對抗網路,逆向強化學習和 Energy-Based 模型之間的聯繫(A Connection between Generative Adversarial Networks, Inverse Reinforcement Learning, and Energy-Based Models )
可擴展對抗分類的通用訓練框架(A General Retraining Framework for Scalable Adversarial Classification)
對抗自編碼器(Adversarial Autoencoders)
對抗判別的領域適應(Adversarial Discriminative Domain Adaptation)
對抗性 Generator-Encoder 網路(Adversarial Generator-Encoder Networks)
對抗特徵學習(Adversarial Feature Learning)
【代碼】https://github.com/wiseodd/generative-models
對抗推理學習(Adversarially Learned Inference)
【代碼】https://github.com/wiseodd/generative-models
結構化輸出神經網路半監督訓練的一種對抗正則化(An Adversarial Regularisation for Semi-Supervised Training of Structured Output Neural Networks)
聯想式對抗網路(Associative Adversarial Networks)
b-GAN:生成對抗網路的一個新框架(b-GAN: New Framework of Generative Adversarial Networks)
【代碼】https://github.com/wiseodd/generative-models
邊界尋找生成對抗網路(Boundary-Seeking Generative Adversarial Networks)
【代碼】https://github.com/wiseodd/generative-models
條件生成對抗網路(Conditional Generative Adversarial Nets)
【代碼】https://github.com/wiseodd/generative-models
結合生成對抗網路和 Actor-Critic 方法(Connecting Generative Adversarial Networks and Actor-Critic Methods)
描述符和生成網路的協同訓練(Cooperative Training of Descriptor and Generator Networks)
Coupled Generative Adversarial Networks(CoGAN)
【代碼】https://github.com/wiseodd/generative-models
基於能量模型的生成對抗網路(Energy-based Generative Adversarial Network)
【代碼】https://github.com/wiseodd/generative-models
對抗樣本的解釋和利用(Explaining and Harnessing Adversarial Examples)
f-GAN:使用變分發散最小化訓練生成式神經採樣器(f-GAN: Training Generative Neural Samplers using Variational Divergence Minimization)
Gang of GANs: Generative Adversarial Networks with Maximum Margin Ranking
用遞歸對抗網路乘車圖像(Generating images with recurrent adversarial networks)
Generative Adversarial Nets with Labeled Data by Activation Maximization
生成對抗網路(Generative Adversarial Networks)
【代碼】https://github.com/goodfeli/adversarial
【代碼】https://github.com/wiseodd/generative-models
生成對抗並行化(Generative Adversarial Parallelization)
【代碼】https://github.com/wiseodd/generative-models
One Shot學習的生成性對抗殘差成對網路(Generative Adversarial Residual Pairwise Networks for One Shot Learning)
生成對抗結構化網路(Generative Adversarial Structured Networks)
生成式矩匹配網路(Generative Moment Matching Networks)
【代碼】https://github.com/yujiali/gmmn
訓練GAN的改進技術(Improved Techniques for Training GANs)
【代碼】https://github.com/openai/improved-gan
改善訓練WGAN(Improved Training of Wasserstein GANs)
【代碼】https://github.com/wiseodd/generative-models
InfoGAN:通過信息最大化GAN學習可解釋表示(InfoGAN: Interpretable Representation Learning by Information Maximizing Generative Adversarial Nets)
【代碼】https://github.com/wiseodd/generative-models
翻轉GAN的生成器(Inverting The Generator Of A Generative Adversarial Network)
隱式生成模型里的學習(Learning in Implicit Generative Models)
用GAN學習發現跨域關係(Learning to Discover Cross-Domain Relations with Generative Adversarial Networks)
【代碼】https://github.com/wiseodd/generative-models
最小二乘生成對抗網路,LSGAN(Least Squares Generative Adversarial Networks)
【代碼】https://github.com/wiseodd/generative-models
LS-GAN,損失敏感GAN(Loss-Sensitive Generative Adversarial Networks on Lipschitz Densities)
LR-GAN:用於圖像生成的分層遞歸GAN(LR-GAN: Layered Recursive Generative Adversarial Networks for Image Generation)
MAGAN: Margin Adaptation for Generative Adversarial Networks
【代碼】https://github.com/wiseodd/generative-models
最大似然增強的離散生成對抗網路(Maximum-Likelihood Augmented Discrete Generative Adversarial Networks)
模式正則化GAN(Mode Regularized Generative Adversarial Networks)
【代碼】https://github.com/wiseodd/generative-models
Multi-Agent Diverse Generative Adversarial Networks
生成對抗網路中Batch Normalization和Weight Normalization的影響(On the effect of Batch Normalization and Weight Normalization in Generative Adversarial Networks)
基於解碼器的生成模型的定量分析(On the Quantitative Analysis of Decoder-Based Generative Models)
SeqGAN:策略漸變的序列生成對抗網路(SeqGAN: Sequence Generative Adversarial Nets with Policy Gradient)
深度網路的簡單黑箱對抗干擾(Simple Black-Box Adversarial Perturbations for Deep Networks)
Stacked GAN(Stacked Generative Adversarial Networks)
通過最大均值差異優化訓練生成神經網路(Training generative neural networks via Maximum Mean Discrepancy optimization)
Triple Generative Adversarial Nets
Unrolled Generative Adversarial Networks
DCGAN無監督表示學習(Unsupervised Representation Learning with Deep Convolutional Generative Adversarial Networks)
【代碼】https://github.com/Newmu/dcgan_code
【代碼】https://github.com/pytorch/examples/tree/master/dcgan
【代碼】https://github.com/carpedm20/DCGAN-tensorflow
【代碼】https://github.com/jacobgil/keras-dcgan
Wasserstein GAN(WGAN)
【代碼】https://github.com/martinarjovsky/WassersteinGAN
【代碼】https://github.com/wiseodd/generative-models
視覺應用
用對抗網路檢測惡性前列腺癌(Adversarial Networks for the Detection of Aggressive Prostate Cancer)
條件對抗自編碼器的年齡遞進/回歸(Age Progression / Regression by Conditional Adversarial Autoencoder)
ArtGAN:條件分類GAN的藝術作品合成(ArtGAN: Artwork Synthesis with Conditional Categorial GANs)
Beyond Face Rotation: Global and Local Perception GAN for Photorealistic and Identity Preserving Frontal View Synthesis
卷積人臉生成的條件GAN(Conditional generative adversarial nets for convolutional face generation)
輔助分類器GAN的條件圖像合成(Conditional Image Synthesis with Auxiliary Classifier GANs)
【代碼】https://github.com/wiseodd/generative-models
使用對抗網路的Laplacian金字塔的深度生成圖像模型(Deep Generative Image Models using a Laplacian Pyramid of Adversarial Networks)
【代碼】https://github.com/facebook/eyescream
【博客】http://soumith.ch/eyescream/
Deep multi-scale video prediction beyond mean square error
【代碼】https://github.com/dyelax/Adversarial_Video_Generation
DualGAN:圖像到圖像翻譯的無監督Dual學習(DualGAN: Unsupervised Dual Learning for Image-to-Image Translation)
【代碼】https://github.com/wiseodd/generative-models
用循環神經網路做全解析度圖像壓縮(Full Resolution Image Compression with Recurrent Neural Networks)
生成以適應:使用GAN對齊域(Generate To Adapt: Aligning Domains using Generative Adversarial Networks)
生成對抗文本到圖像的合成(Generative Adversarial Text to Image Synthesis)
【代碼】https://github.com/paarthneekhara/text-to-image
自然圖像流形上的生成視覺操作(Generative Visual Manipulation on the Natural Image Manifold)
【項目】http://www.eecs.berkeley.edu/~junyanz/projects/gvm/
【視頻】https://youtu.be/9c4z6YsBGQ0
【代碼】https://github.com/junyanz/iGAN
Image De-raining Using a Conditional Generative Adversarial Network
Image Generation and Editing with Variational Info Generative Adversarial Networks
用條件對抗網路做 Image-to-Image 翻譯(Image-to-Image Translation with Conditional Adversarial Networks)
【代碼】https://github.com/phillipi/pix2pix
用GAN模仿駕駛員行為(Imitating Driver Behavior with Generative Adversarial Networks)
可逆的條件GAN用於圖像編輯(Invertible Conditional GANs for image editing)
學習驅動模擬器(Learning a Driving Simulator)
多視角GAN(Multi-view Generative Adversarial Networks)
利用內省對抗網路編輯圖片(Neural Photo Editing with Introspective Adversarial Networks)
使用GAN生成照片級真實感的單一圖像超解析度(Photo-Realistic Single Image Super-Resolution Using a Generative Adversarial Network)
Recurrent Topic-Transition GAN for Visual Paragraph Generation
RenderGAN:生成現實的標籤數據(RenderGAN: Generating Realistic Labeled Data)
SeGAN: Segmenting and Generating the Invisible
使用對抗網路做語義分割(Semantic Segmentation using Adversarial Networks)
半隱性GAN:學習從特徵生成和修改人臉圖像(Semi-Latent GAN: Learning to generate and modify facial images from attributes)
TAC-GAN - 文本條件輔助分類器GAN(TAC-GAN - Text Conditioned Auxiliary Classifier Generative Adversarial Network)
通過條件GAN實現多樣化且自然的圖像描述(Towards Diverse and Natural Image Descriptions via a Conditional GAN)
GAN 提高人的體外識別基線的未標記樣本生成(Unlabeled Samples Generated by GAN Improve the Person Re-identification Baseline in vitro)
Unpaired Image-to-Image Translation using Cycle-Consistent Adversarial Networks
無監督異常檢測,用GAN指導標記發現(Unsupervised Anomaly Detection with Generative Adversarial Networks to Guide Marker Discovery)
無監督跨領域圖像生成(Unsupervised Cross-Domain Image Generation)
WaterGAN:實現單目水下圖像實時顏色校正的無監督生成網路(WaterGAN: Unsupervised Generative Network to Enable Real-time Color Correction of Monocular Underwater Images)
其他應用
基於生成模型的文本分類的半監督學習方法(Adversarial Training Methods for Semi-Supervised Text Classification)
學習在面對對抗性神經網路解密下維護溝通保密性(Learning to Protect Communications with Adversarial Neural Cryptography)
【博客】http://t.cn/RJitWNw
MidiNet:利用 1D 和 2D條件實現符號域音樂生成的卷積生成網路(MidiNet: A Convolutional Generative Adversarial Network for Symbolic-domain Music Generation using 1D and 2D Conditions)
使用生成對抗網路重建三維多孔介質(Reconstruction of three-dimensional porous media using generative adversarial neural networks)
【代碼】https://github.com/LukasMosser/PorousMediaGan
Semi-supervised Learning of Compact Document Representations with Deep Networks
Steganographic GAN(Steganographic Generative Adversarial Networks)
Humor
視頻
Ian Goodfellow:生成對抗網路
【視頻】http://t.cn/RxxJF5A
Mark Chang:生成對抗網路教程
【視頻】http://t.cn/RXJOKK1
代碼
Cleverhans:一個對抗樣本的機器學習庫
【代碼】https://github.com/openai/cleverhans
【博客】http://cleverhans.io/
50行代碼實現GAN(PyTorch)
【代碼】https://github.com/devnag/pytorch-generative-adversarial-networks
【博客】https://medium.com/@devnag/generative-adversarial-networks-gans-in-50-lines-of-code-pytorch-e81b79659e3f
生成模型集,e.g. GAN, VAE,用 Pytorch 和 TensorFlow 實現
【代碼】https://github.com/wiseodd/generative-models
【進入新智元微信公眾號,在對話框輸入「170501」下載全部 89 篇論文】
原文地址:https://github.com/nightrome/really-awesome-gan
新智元招聘
職位:客戶總監
職位年薪:30 - 60萬(工資+獎金)
工作地點:北京-海淀區
所屬部門:客戶部
彙報對象:COO
下屬人數:8 人
工作年限:5 年
語 言:英語 + 普通話
學歷要求:全日制統招本科
職位描述:
熱愛人工智慧,在行業內有一定的人脈資源和影響力;
為客戶制定媒體關係策略和公關活動策劃,達成客戶的市場或傳播目標;
負責監督公關項目的計劃和實施,使項目能按期在預算內完成;
積極拓展客戶資源,開發公司業務,與既有客戶保持緊密的業務聯絡和溝通;
監督、管理及考核客戶服務團隊,全面提升公司客戶服務質量;
有4A、公關公司工作經歷優先,有知名企業或知名媒體機構工作經驗者加分。
崗位職責:
精準把握客戶需求,了解新智元定位,負責制定、執行公司年度會展活動及其他客戶經營服務計劃,監督實施全過程,完成業績指標。掌握 AI 產業動態,熟悉市場狀況,定期提出客戶服務計劃和建議。積極開拓新客戶,提升與既有客戶、同行業間良好的合作關係;組織部門開發多種服務計劃,完成客戶服務目標及確保回款任務。負責管理和培訓客戶部業務人員,安排並指導客戶部日常工作,與其他部門良好協作。
應聘郵箱:jobs@aiera.com.cn
HR微信:13552313024
新智元歡迎有志之士前來面試,更多招聘崗位請點擊【新智元招聘】查看。
![](https://pic.pimg.tw/zzuyanan/1488615166-1259157397.png)
![](https://pic.pimg.tw/zzuyanan/1482887990-2595557020.jpg)
※讓神經網路替你編程:如何用深度學習實現程序自動合成
※量產在即:特斯拉Model3細節曝光,ModelS和X評價降級
※「一圖看懂」計算機視覺識別簡史:從 AlexNet、ResNet 到 Mask RCNN
TAG:新智元 |
※文字、視頻、VR、AR NBA直播的過去、現在和將來
※【下載】61個0-12歲分齡動畫、電影、閱讀、英語、視頻音頻全套下載!
※美圖AAAI 2018論文:視頻語義理解的類腦智能
※3D視頻拍攝硬體及演算法,生成實時互動&實時分享的3D視頻內容
※OPPO展示5G驚人技術:7秒輕鬆下完1G視頻;網友:話費又該不夠了
※1800G資源之色彩運用基礎(下)視頻教程
※GPDR條例和視頻檢索論文閱讀
※影之詩Steam版下載視頻教程
※100G-UI視頻資源和圖標
※1800G資源之C4D實例視頻教程
※[視頻]HTC U12原型上手:實測下行速度可達809.58Mbps
※OPPO實現3D結構光5G視頻通話技術,首個5G視頻通話將實現商用
※DeepMind的腦補AI再獲新技能:看文字知場景、複雜環境、連續視頻……
※更細分和更個性化的電視和視頻業態成為事實,2018年電視和OTT同義
※OPPO的黑科技?基於5G的3D視頻通話技術曝光
※7大實戰案例、24個學習視頻、12G乾貨資料...帶你免費入門「Python數據分析」
※原價45元的語音輸字幕視頻編輯工具等6款App限免#iOS
※OPPO再展示黑科技:基於5G的3D視頻通話技術曝光
※LG新旗艦LG G7 Neo渲染視頻和諜照曝光 5月發布
※YouTube刪除數百個宣傳論文代寫網站的視頻