Deep generative models (DGMs) of images are now sufficiently mature that they produce nearly photorealistic samples and obtain scores similar to the data distribution on heuristics such as Frechet Inception Distance (FID). These results, especially on large-scale datasets such as ImageNet, suggest that DGMs are learning the data distribution in a perceptually meaningful space and can be used in downstream tasks. To test this latter hypothesis, we use class-conditional generative models from a number of model classes---variational autoencoders, autoregressive models, and generative adversarial networks (GANs)---to infer the class labels of real data. We perform this inference by training an image classifier using only synthetic data and using the classifier to predict labels on real data. The performance on this task, which we call Classification Accuracy Score (CAS), reveals some surprising results not identified by traditional metrics and constitute our contributions. First, when using a state-of-the-art GAN (BigGAN-deep), Top-1 and Top-5 accuracy decrease by 27.9\% and 41.6\%, respectively, compared to the original data; and conditional generative models from other model classes, such as Vector-Quantized Variational Autoencoder-2 (VQ-VAE-2) and Hierarchical Autoregressive Models (HAMs), substantially outperform GANs on this benchmark. Second, CAS automatically surfaces particular classes for which generative models failed to capture the data distribution, and were previously unknown in the literature. Third, we find traditional GAN metrics such as Inception Score (IS) and FID neither predictive of CAS nor useful when evaluating non-GAN models. Furthermore, in order to facilitate better diagnoses of generative models, we open-source the proposed metric.
翻译:图像的深基因模型(DGM)现在已经足够成熟,它们制作了近光现实样本,并获得了类似于Frechet Inpeption Contreal(FID)等超光层数据分布的分数。这些结果,特别是图像Net(FID)等大型数据集的结果,表明DGM正在一个感知有意义的空间里学习数据分布,可用于下游任务。为了测试后一种假设,我们使用了从一些模型级变异自动图解样本、自动反向模型和基因对抗网络(GANs)获得类似分数的分数。我们通过培训仅使用合成数据并使用分类来预测真实数据标签的图像分类,表明DGGGGM(GAN-2)的分数、自动递增模型(GAN-2)的分数,Top-1和Top-5等类类数据。我们使用SDVARS)的分数显示一些令人惊讶的结果,在使用州-Art GAN(BAAN(BAN-I-I) 的分数级、OAD-Salal-dealal-dealalalalalalalal delal-deal-deal-deal-deal deal demodeal demodeal demode dal deal demodal demodal demod dism dal deal demodal demod dism dism dism dismal dis deal)中, 和这种数据,在Smal-de dism disal-de dismaldal demode dismodal demodaldaldaldaldald dismod 或27.9-de dismodal-de dismod-de dismodal-de dismod-de dism dism dism dism dism dism dism dismaldaldaldaldaldaldaldald saldaldaldaldald 和制数据中找不到数据中找不到数据中找不到数据中找不到数据中找不到数据中,在比比比第4,比比第4,比第4,比第4,比第4,比第4,比第