GANs Trained by a Two Time-Scale Update Rule Converge to a Local Nash Equilibrium

By Martin Heusel, Hubert Ramsauer, Thomas Unterthiner, Bernhard Nessler, Sepp Hochreiter. 2017

18,997 citations recorded by the source corpus.

Subfields: Generative Models, Optimization & Training

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Generative Models

In Optimization & Training

Citations in the corpus

Denoising Diffusion Probabilistic Models (2020, methodology): “On CIFAR10, we calculated Inception and FID scores on 50000 samples using the original code from the OpenAI [48] and TTUR [19] repositories, respectively.”
High-Resolution Image Synthesis with Latent Diffusion Models (2021, methodology): “6, we compare models trained on CelebAHQ [38] and ImageNet in terms sampling speed for different numbers of denoising steps with the DDIM sampler [81] and plot it against FID-scores [28].”
Hierarchical Text-Conditional Image Generation with CLIP Latents (2022, methodology): “Finally, in Figure 10 we compute MS-COCO zero-shot FID [23] while sweeping over guidance scale for both unCLIP and GLIDE, finding that guidance hurts the FID of unCLIP much less so than for GLIDE.”
A Large-scale Study of Representation Learning with the Visual Task Adaptation Benchmark (2019, background): “Generative Evaluation Evaluation of generative quality is a difficult goal with numerous proxy metrics, such as reconstruction error, FID (Heusel et al., 2017), IS (Salimans et al.”
Scaling Autoregressive Video Models (2019, methodology): “This is the spatiotemporal counterpart to the Fréchet Inception Distance [16], replacing the ImageNet-trained Inception network of the latter with an I3D Network trained on Kinetics.”
Natural Adversarial Examples (2019, positive): “The Fréchet Inception Distance (FID) [31] enables us to de-”
Are GANs Created Equal? A Large-Scale Study (2017, methodology): “Fréchet Inception Distance is computed by considering the difference in embedding of true and fake data [11].”
Zero-Shot Text-to-Image Generation (2021, methodology): “We evaluate our model zero-shot by comparing it to three prior approaches: AttnGAN (Xu et al., 2018), DMGAN (Zhu et al., 2019), and DF-GAN (Tao et al., 2020), the last of which reports the best Inception Score (Salimans et al., 2016) and Fréchet Inception Distance (Heusel et al., 2017) on MS-COCO.”
Evaluating Lossy Compression Rates of Deep Generative Models (2020, background): “One of the main drawbacks of the IS or FID is that they can only provide a single scalar value that cannot distinguish the mode dropping behavior from the mode inventing behavior in generative models.”
Improved Techniques for Training Score-Based Generative Models (2020, methodology): “[9] noticed that adding an extra denoising step after the original annealed Langevin dynamics in [1], similar to [10, 11, 12], often significantly improves FID scores [13] without affecting the visual appearance of samples.”
Training Generative Adversarial Networks with Limited Data (2020, methodology): “The bolded numbers indicate the lowest FID and highest Recall for each training set size.”
Your Classifier is Secretly an Energy Based Model and You Should Treat it Like One (2019, methodology): “First we note that both IS and FID are scores computed based on a pretrained classifier network, and thus can be very dependent on the exact model/code repository used.”
Flow Contrastive Estimation of Energy-Based Models (2019, methodology): “To evaluate the ﬁdelity of synthesized examples, Table 1 summarizes the Fr´echet Inception Distance (FID) [24] of the synthesized examples computed with the Inception V3 [65] classiﬁer.”
Generative Modeling by Estimating Gradients of the Data Distribution (2019, methodology): “1, we compute inception and FID scores based on a total of 50000 samples.”
Generating Diverse High-Fidelity Images with VQ-VAE-2 (2019, methodology): “The two most common metrics for comparing GANs are Inception Score [34] and Fréchet Inception Distance (FID) [13].”
Learning Non-Convergent Non-Persistent Short-Run MCMC Toward Energy-Based Model (2019, background): “Table 1 (a) compares the Inception Score (IS) [45, 4] and Fréchet Inception Distance (FID) [20] with Inception v3 classiﬁer [47] on 40 , 000 generated examples.”
A Style-Based Generator Architecture for Generative Adversarial Networks (2018, positive): “Table 1 gives Fréchet inception distances (FID) [24] for various generator architectures in CelebA-HQ [28] and our new FFHQ dataset (Appendix A).”
Large Scale GAN Training for High Fidelity Natural Image Synthesis (2018, background): “A variety of works have proposed heuristics for measuring the sample quality of models without tractable likelihoods (Salimans et al., 2016; Heusel et al., 2017; Bińkowski et al., 2018; Wu et al., 2017).”