Unsupervised Representation Learning with Deep Convolutional Generative Adversarial Networks

19 Nov 2015Alec RadfordLuke MetzSoumith Chintala

In recent years, supervised learning with convolutional networks (CNNs) has seen huge adoption in computer vision applications. Comparatively, unsupervised learning with CNNs has received less attention... (read more)

PDF Abstract

Code


Results from the Paper


TASK DATASET MODEL METRIC NAME METRIC VALUE GLOBAL RANK RESULT BENCHMARK
Image Classification CIFAR-10 DCGAN Percentage correct 82.8 # 81
Conditional Image Generation CIFAR-10 DCGAN Inception score 6.58 # 12
Image Classification CIFAR-10 1 Layer K-means Percentage correct 80.6 # 84
Image Classification SVHN TSVM Percentage error 66.55 # 39
Image Classification SVHN Supervised CNN Percentage error 28.87 # 35
Image Classification SVHN KNN Percentage error 77.93 # 40
Image Classification SVHN DCGAN Percentage error 22.48 # 33

Results from Other Papers


TASK DATASET MODEL METRIC NAME METRIC VALUE GLOBAL RANK SOURCE PAPER COMPARE
Image Clustering CIFAR-10 GAN Accuracy 0.315 # 4
NMI 0.265 # 4
Image Clustering CIFAR-100 GAN Accuracy 0.151 # 5
NMI 0.120 # 4
Image Clustering ImageNet-10 GAN Accuracy 0.346 # 4
NMI 0.225 # 4
Image Clustering Imagenet-dog-15 GAN Accuracy 0.174 # 5
NMI 0.121 # 4
Image Clustering STL-10 GAN Accuracy 0.298 # 5
NMI 0.210 # 5
Image Clustering Tiny-ImageNet GAN Accuracy 0.041 # 3
NMI 0.135 # 3

Methods used in the Paper