Generative adversarial network (GAN)

Overview

A GAN trains two networks at once. The discriminator D learns to tell real samples from generated ones, and the generator G learns to produce samples D accepts as real. The baseline objective is a max over D of the expected log score on real samples plus the log of one minus the score on generated ones, with G minimising the same quantity 1611.02200 §3, eq. 1 1710.07346 §3, eq. 1.

Eighteen cards train a generator against a discriminator or critic: unpaired translation, restyling of labelled aerial images, paired MR-to-CT synthesis, two-stage fashion synthesis, multi-domain translation with style codes, stacked text-to-image synthesis, motion generation, long text generation, sequence text generation by policy gradient, a latent GAN for point-cloud completion, a conditional GAN with trained encoders for face editing, a transformer generator for underwater enhancement, one tabular GAN per failure type, a recurrent GAN for continuous music events, a residual GAN for MRI reconstruction, a patch GAN trained on one image to estimate its blur kernel, a class-conditional GAN for source-free adaptation, and a refinement GAN that fills the holes left by a point-cloud renderer for single-image view synthesis. Most of them condition G on an input or a label rather than on noise alone. The fashion model also takes noise 1710.07346 §3.1. The music card's generator takes noise at every step and no image or label input 1611.09904 §2 §3. The kernel card's generator takes the image itself, and the image is the only data it sees 1909.06581 §2 §5.3. The text generator in the long-text card has no input beyond a start token and its own prefix 1709.08624 §Methodology. The SeqGAN generator has the same kind of input, and its discriminator scores whole sequences only 1609.05473 §Method. The latent GAN in the shape card takes only a one-dimensional seed, which a separate agent chooses from the scan 1904.12304 §3.2 §3.3. Of the five translation-style cards, four put the adversarial term beside reconstruction, identity, cycle or style terms. The fourth is StarGAN v2 1912.01865 eq. 5. The fashion card gives no extra term beyond the GAN objective 1710.07346 §3.4. The motion card weights the adversarial term at 0.1 against its reconstruction losses 2102.04942 §3.7.3. The text-to-image card adds a contrastive term to the adversarial losses it inherits from AttnGAN 1904.01480 §3.1 §4.1. The underwater card adds its adversarial term with no coefficient, so weight one, beside LAB, LCH, RGB and perceptual terms 2111.11843 eq. 13. The IcGAN card trains its encoders with squared losses and no adversarial term 1611.06355 eq. 3 eq. 4. The tabular card trains each generator as a WGAN-GP and then filters its samples with a separate classifier 2607.19153 §3.3.1.

Three more cards also train a discriminator. The age-progression card puts discriminators on both its latent code and its output image 1702.08423 eq. 3-4. The distillation card trains a student against a diffusion teacher, with a discriminator on frozen features of the student's output 2311.17042 §3.2. The video card judges whole clips: its discriminator reads k sampled frames of one clip, three by default, and scores them jointly 2112.14683 §3.2. All three are described below.

Six more cards use GANs differently. The minimax transport card has a two-player game, but its inner player is a convex potential, not a real-or-fake classifier 1908.10962 §3. The StyleFlow and VQGAN+CLIP cards reuse a pretrained GAN generator and train no discriminator 2008.02401 §6.2 2204.08583 §2. The StyleGAN-NADA card also trains no discriminator, but it does update a pretrained StyleGAN, with a CLIP-based directional loss as its only training signal 2108.00946 §4.1. Two inversion cards also use a pretrained StyleGAN without a discriminator. The Image2StyleGAN++ card keeps the generator frozen and optimises a W+ code and then per-layer noise maps for one photo 1911.11544 §5 §4.2. The PTI card freezes the W code of one face, then fine-tunes the whole generator so that it reproduces that face, with a locality penalty on nearby codes 2106.05744 §3.2 §3.3. The two cards therefore differ in whether the generator's weights change.

Two more cards fit a pretrained GAN with no discriminator of their own. The 3D face card optimises a 3D face model and a texture latent against one photo, through the GAN's output 1902.05978 §3.1. The point-editing card optimises a StyleGAN2 latent so that handle points move to their targets 2305.10973 §3.2. Both are described below.

Nine cards use a GAN without making it the method. Texygen is a benchmark harness. It packages seven text generators, six of them GAN variants and one maximum-likelihood baseline, with five metrics 1802.01886 §2.1 §2.2. WaveFake builds a detection dataset whose fake speech comes from several generators, some of them GAN-based vocoders, and its detectors are statistical models and a CNN-GRU, not GANs 2111.02813 §2.2 §3. The EEG survey lists GAN decoders among its image decoders, next to VAE and latent diffusion, and reports no new results of its own 2502.12048 §3 §4.5. The deep-learning survey has one section on GANs, which gives the minimax objective and the usual remark that the generator loss can saturate early. Its printed non-saturating term omits the discriminator, so take the exact form from the original GAN paper 1803.01164 §VII.A. The generative-model survey covers energy-based, variational, adversarial, autoregressive and flow models, and it compares them in one star-rated table on CIFAR-10. Its FID and NLL values come from the cited works, and it lists BigGAN at 14.73 and StyleGAN2 with ADA at 2.42 2103.04922 tab. 1. The table is not like-for-like, and its GAN loss rows mix losses to minimise with losses to maximise, so take exact forms from the original papers 2103.04922 fig. 5a tab. 1. The augmentation benchmark compares nine augmenters, one of them CT-GAN, with nine ensemble models on binary and multi-class imbalanced sets 2304.02858 §4.1 §5.1. The CT-GAN row of its table is one arm of that comparison, not a method the paper proposes 2304.02858 tab. 1. The social navigation card uses a Social GAN generator as a pedestrian forecaster inside a robot planner. Its own contribution is extra cost terms for the planner, and the paper does not say which data the forecaster was trained on 2411.01814 §2.1 eq. 5–8. Two reviews name GANs among several generative models. The graph-neural-network review describes graph GANs as one kind of graph generative network, with a min-max loss over node pairs. It does not give their architectures, and it quotes one synthetic-feeder comparison in which they reach a 27.2% connected rate, against 8.3% for a random-matrix method. The dataset is not named 2101.10025 §II-E §III-D2 tab. VI. The GAN-evaluation review updates the measures used for GANs, from FID variants to precision-recall extensions and human judgement. It runs no experiments and quotes its numbers from other papers 2103.09396 §2 §Results.

One card is an evaluation metric rather than a method. The Fréchet Video Distance card does not train a GAN. It compares Gaussian fits of video features for generators in general, and GANs are cited there as models that lack ground truth for a direct comparison 1812.01717 §2. Identical sets have a nonzero distance, so the card says to compare only at equal sample counts 1812.01717 §4.2.

Another card judges generative models for engineering design. It sorts its metrics into similarity, novelty and diversity, constraint, performance and conditioning groups 2302.02913 §Method. Its running example is one GAN and one VAE on six-mode 2-D data 2302.02913 §4.5. There the GAN has the lower maximum mean discrepancy, 0.013 against 0.045, and the higher constraint satisfaction, 0.833 against 0.713. The VAE has the higher precision-recall F1, 0.475 against 0.386 2302.02913 tab. 2 tab. 4. The paper runs one instance per model and tests no difference for significance §Pitfalls. The same nearest-datapoint score is minimised as a realism measure and maximised as a novelty measure, so the direction depends on the question [arxiv:2302.02913 tab. 1] §5.1. The released code covers the 2-D examples only §Paper vs code.

Two more cards use a GAN as one part of a larger system. The partial multi-view card's core model, CPM-Nets, has no adversarial term. Its CPM-GAN variant adds one discriminator per view, trained on observed values as positives, to fill in missing entries 2011.06170 §3 alg. 2. The released code has no GAN part 2011.06170 §Paper vs code. The emotional speech card is mainly a dataset paper. It benchmarks two of its own converters, CycleGAN-EVC and VAWGAN-EVC, on one speaker 2105.14762 §5.1. The GAN is a baseline there, not the contribution 2105.14762 §2 §3.

The numbers below are author-reported, on different data and with different metrics. The page therefore compares design choices rather than one benchmark.

Variants

Unpaired translation with a frozen feature function (DTN)

The Domain Transfer Network learns a map from a source domain to a target domain using unpaired samples from both. A fixed, pretrained function f supplies the representation. G is built as a small trainable network g applied after f, so only g decodes f's features 1611.02200 §4. The discriminator has three classes: generated from source, generated from target, and real target. The generator must push both generated classes towards the real-target class 1611.02200 eq. 3, eq. 4. Two further terms keep the output faithful. One asks f to give the same features before and after translation for source samples. The other asks G to leave target samples unchanged 1611.02200 eq. 5, eq. 6. A total-variation smoothness term is also defined, but the digit setting sets its weight to zero [arxiv:1611.02200 §5.1]. Card: Unsupervised Cross-Domain Image Generation.

On SVHN to MNIST, the full model reaches 90.66%. A plain binary GAN with feature constancy reaches 13.71% in the same table. Removing the identity term gives 88.40%, and removing the constancy term gives 74.55% 1611.02200 tab. 1. On faces, translated emoji give a median rank of 16 for identity retrieval, against 16311 for hand-made emoji. The face experiment uses a non-public face set and a non-public feature network 1611.02200 tab. 4, §5.2.

Cycle-consistent restyling of labelled source images (aerial segmentation)

The aerial card adapts a segmenter trained on one labelled aerial dataset to another, with no target labels. It trains a translator in both directions on unpaired patches from the two datasets. Each direction has its own adversarial loss, and a reconstruction term asks each round trip to return its input 1905.03198 §4.1 eq. 3 eq. 8. The source labels are then reused for the translated source images, and the source segmenter is fine-tuned on them 1905.03198 §4.2 §4.3. The segmenter is BiSeNet with a ResNet-101 feature extractor 1905.03198 §5.2.1. Card: Unsupervised Domain Adaptation using Generative Adversarial Networks for Semantic Segmentation of Aerial Images.

On Potsdam to Vaihingen, average accuracy goes from 0.35 before adaptation to 0.52 after 1905.03198 tab. 3. Building accuracy goes from 0.23 to 0.71 and Tree from 0.06 to 0.51. Low vegetation falls from 0.38 to 0.27 1905.03198 tab. 4. The discussion calls the other classes unchanged 1905.03198 §6, which the Low vegetation row does not support. The target model is scored on the Vaihingen test set after every epoch, and the reported gain is taken after eight epochs 1905.03198 §5.2.4. That choice uses target labels, so the best-epoch numbers are optimistic. The authors' repository has no license file 1905.03198.

Paired 3D conditional GAN with auto-context (MR to CT)

The second variant is supervised. A 3D fully convolutional generator maps a 32³ MRI patch to the matching 16³ CT patch, and a 3D CNN discriminator judges real and generated CT patches 1612.05362 §2.1, §4. The generator loss combines a weighted adversarial term, an L2 term against the true CT, and a 3D gradient-difference term. The weights are 0.5, 1 and 1 [arxiv:1612.05362 eq. 5, eq. 7] 1612.05362 §4. A stack of such GANs is trained in sequence with the Auto-Context Model. Each stage also receives the previous stage's CT estimate as an extra input channel. The paper uses two such stages 1612.05362 §2.2, §4.2. Card: Medical Image Synthesis with Context-Aware Generative Adversarial Networks.

On 16 brain and 22 pelvis subjects, under leave-one-out evaluation, the author-reported PSNR is 27.6 for brain and 34.1 for pelvis. The strongest baseline, structured random forest with auto-context, reports 26.3 and 32.1 1612.05362 tab. 1, tab. 2. The baseline numbers are copied from that method's paper rather than re-run 1612.05362 §4. In an ablation where the dataset is not stated, adding the adversarial term raises PSNR from 24.7 to 25.9 over a generator trained alone 1612.05362 §4.1.

Paired transformer generator with colour-space losses (underwater enhancement)

The U-shape Transformer keeps the paired image-to-image layout of pix2pix, an encoder, a bottleneck and a decoder, and changes two parts. A channel-wise attention block (CMSFFT) replaces the skip connections. It attends across colour channels, using the four encoder scales as its inputs [arxiv:2111.11843 §3.2.3] eq. 4 eq. 7. A four-layer spatial transformer (SGFMT) sits at the bottleneck 2111.11843 §3.2.2 eq. 1 eq. 3. The generator loss adds LAB and LCH colour losses, an RGB L2 term and a VGG perceptual term to a GAN term 2111.11843 eq. 13. Its LSUI dataset has 4279 pairs of underwater photos and reference images, and 3879 of them are used for training 2111.11843 §3.1 §4.1. Card: U-shape Transformer for Underwater Image Enhancement.

Author-reported PSNR on Test-L400 is 24.16, against 22.91 for the next best method, Ucolor, with SSIM 0.93 against 0.89 2111.11843 tab. 3. On Test-U90 the PSNR gap is 2.13 dB, and on Test-L400 it is 1.25 dB, so the abstract's "more than 2 dB" holds only for one of the two sets [arxiv:2111.11843 tab. 3]. Adding the channel block to a pix2pix baseline raises Test-L400 PSNR from 19.34 to 22.47, and the spatial block raises it to 21.78 [arxiv:2111.11843 tab. 6]. The ablation has no run without the adversarial term, so the paper does not isolate what the GAN contributes 2111.11843 tab. 6. The paper says the source photos were licensed for academic use only, so check that before training on LSUI for a commercial product 2111.11843 §3.1.

One generator with style codes for many domains (StarGAN v2)

StarGAN v2 uses one generator G(x, s) for all domains, so it does not need one translator per pair of domains. The style code s comes either from a mapping network, which reads random noise and a target domain, or from a style encoder, which reads a reference image and its domain. Both have one output head per domain, so every domain has its own style space 1912.01865 §2.1 [app. E]. The generator injects the style into each decoder block through adaptive instance normalization 1912.01865 §2.1. The discriminator has one real-or-fake output per domain 1912.01865 §2.1.

Training adds three terms to the adversarial loss 1912.01865 eq. 5. A style-reconstruction term asks the encoder to recover the style that was used eq. 2. A diversity term asks two different style codes to give different images eq. 3. A cycle term translates the output back with the input's own estimated style, so pose and layout survive eq. 4. The paper drops the denominator from the original diversity-sensitive loss, which it says made training unstable, and it decays the diversity weight linearly to zero 1912.01865 §2.2 app. B. Card: StarGAN v2: Diverse Image Synthesis for Multiple Domains.

On CelebA-HQ, the author-reported latent-guided FID is 13.7, against 33.1 for MSGAN and 31.4 for MUNIT 1912.01865 tab. 2. On AFHQ it is 16.2, against 61.4 for MSGAN tab. 2. Reference-guided FIDs are 23.8 on CelebA-HQ and 19.8 on AFHQ, against 39.6 and 69.8 for MSGAN 1912.01865 tab. 3. In the ablation, the style code and diversity term bring the CelebA-HQ FID from 98.4 for StarGAN to 13.7 1912.01865 tab. 1. In a human study, 68.9% of votes for quality went to StarGAN v2 on CelebA-HQ and 88.1% on AFHQ 1912.01865 tab. 4. The study has 76 valid workers and no significance test 1912.01865 §3.2. The paper's real-image FID on CelebA-HQ is 14.8, and StarGAN v2's 13.7 is below it. The card warns against reading that as beating real images, because the test set is small 1912.01865 tab. 2.

The official code is CC BY-NC 4.0, which rules out commercial use. Its loss uses a mean where the paper writes an L1 norm, so the λ values need the same convention to keep their balance 1912.01865 §Paper vs code. Implement from the paper, and check the AFHQ image licences yourself, because the paper lists none per image 1912.01865 §Pitfalls.

Two-stage conditional GAN through a label map (text-guided fashion)

The fashion model redresses a photo of a person from a text description and keeps the wearer's pose and body shape. It is two conditional GANs trained in sequence. The first generates a 7-label segmentation map. Its input is a coarse 8×8 map of the wearer in which all clothing is merged into one class, plus a design code built from attributes and the text [arxiv:1710.07346 §3.1, §3.2, eq. 2]. The second generates one image channel per label, and each pixel takes its value from the channel of its label [arxiv:1710.07346 eq. 3, eq. 4, §3.3]. The paper's motivation is that plain text-conditioned GANs lose the wearer's structure and blur region boundaries 1710.07346 §1. The two stages are trained separately, because the argmax between them is not differentiable. Both discriminators also receive the conditions 1710.07346 §3.4. Training needs one photo per person, with one description and its segmentation map. No pairs of the same person in two outfits are used 1710.07346 §3.1. Card: Be Your Own Prada: Fashion Synthesis with Structural Coherence.

The author-reported attribute mAP is 82.6, against 71.8 to 74.7 for the baselines and 86.8 on real images 1710.07346 tab. 1. A two-stage model without the per-label mapping reaches 74.7 in the same table, which isolates the mapping's contribution. In a 50-person user study, the full method has the best mean rank 1710.07346 tab. 2. The paper reports no FID, and its user study has no significance test 1710.07346 §4.3.

Conditional autoencoder with discriminators on code and image (age progression)

The age-progression model is a conditional autoencoder, not a plain GAN. An encoder maps a face to a personality vector z. A generator takes z and a 10-bin age label and produces the image at that age. Two discriminators are trained alongside it. One pushes the distribution of z towards a uniform prior, and the other judges generated images conditioned on the age label 1702.08423 eq. 3-4. No paired examples of the same person at two ages are needed 1702.08423 §1. Card: Age Progression/Regression by Conditional Adversarial Autoencoder.

The training set is about 10,670 images 1702.08423 §5.1. The loss weights are λ = 100 and γ = 10 1702.08423 §5.2. The author-reported user preference is 52.77% for this method against 28.99% for prior work 1702.08423 §5.3, and the paper has no face-recognition measure of identity. The paper's vote counts exceed the arithmetic maximum for its panel, so read the percentages with care 1702.08423 §5.3. The paper gives its reconstruction term as L2, but the released code uses L1 1702.08423 §5.2. The total-variation formula, the adversarial weights and the weight initialisation are not stated.

Stacked text-to-image GAN with a Siamese contrastive term (SD-GAN)

SD-GAN is a text-to-image GAN with three generator and discriminator stages. Each caption of a pair goes through its own branch. At every stage, a contrastive loss pulls the two discriminator feature vectors together when the captions describe the same image and pushes them apart when they describe different images 1904.01480 §3.1 §4.3. The pull stops at a floor distance and the push at a margin 1904.01480 eq. 2. The generator's batch norm layers are replaced by semantic-conditioned batch norm (SCBN). Its scale and shift are learned values plus a term that comes either from the sentence vector or from word-level attention 1904.01480 §3.2 eq. 4 eq. 5 eq. 6. Card: Semantics Disentangling for Text-to-Image Generation.

The author-reported Inception score is 4.67 on CUB, against 4.36 for AttnGAN, and 35.69 on MS-COCO, against 25.89 1904.01480 tab. 1. Each part helps on its own. Without the Siamese term the CUB score is 4.49, without SCBN it is 4.51, and without both it is 4.11 1904.01480 tab. 4. In a 50-rater ranking study, SD-GAN was placed first in 68.76% of CUB rankings and 75.78% of COCO rankings, against 20.54% and 17.69% for AttnGAN 1904.01480 tab. 2. The paper tried four values of the floor and 0.1 scored best on both datasets 1904.01480 tab. 5. Its own tables disagree on one setting: a no-Siamese, SCBN-with-word-cue run is 4.49 on CUB in one table and 4.45 in another 1904.01480 tab. 4 tab. 6. No official code was found. The optimiser, learning rate, batch size and contrastive weight are not stated, and the adversarial losses are inherited from AttnGAN 1904.01480 §4.1.

Least-squares critics over motion sequences (in-betweening)

The motion model fills the frames between seed frames and a target keyframe. An LSTM generator produces the transition. Two feed-forward critics score 10-frame windows for long-term structure and 2-frame windows for short-term structure 2102.04942 §3.5. They are trained with a least-squares objective 2102.04942 §3.6.2. The generator's adversarial term has weight 0.1 beside L1 reconstruction losses 2102.04942 §3.7.3. The conditioning relies on two additive embeddings: a sinusoidal time-to-arrival code and a target noise that is switched off within 5 frames of the target 2102.04942 §3.3, §3.4. Card: Robust Motion In-betweening.

The adversarial term is the smallest of the parts. On Human3.6M walking, the average L2Q goes from 0.84 to 0.82 when it is added, and NPSS goes from 0.46 to 0.40 2102.04942 tab. 2. On LaFAN1, the reconstruction-only generator is worse than interpolation on L2P and NPSS at 45 frames 2102.04942 tab. 3. Inference takes about 0.30 s per transition on a CPU, and the plugin uses 170 MB 2102.04942 tab. 4. Training runs for 55,000 iterations on Human3.6M and 350,000 on LaFAN1 2102.04942 §4.2, §4.3.

Sequence generator trained by policy gradient with roll-out rewards (SeqGAN)

SeqGAN treats a text generator as a reinforcement-learning policy. The generator is an LSTM with a softmax output, and each next token is an action taken given the prefix 1609.05473 §Method. The discriminator is a CNN with several kernel widths, max-over-time pooling and one highway layer. It outputs the probability that a finished sequence is real [arxiv:1609.05473 §Method]. The only reward comes at the end. For an earlier token, its value is the average discriminator score over N roll-outs that keep the prefix and finish the sequence with a roll-out policy 1609.05473 eq. 3 eq. 4. The generator then takes a plain gradient-ascent step on the policy gradient 1609.05473 eq. 8. The discriminator is trained by cross entropy between real sequences and generated ones 1609.05473 eq. 5. Training starts with maximum-likelihood pretraining of the generator. Each iteration then takes g generator steps and d discriminator steps, and the discriminator's negatives are regenerated at every step 1609.05473 alg. 1. Card: SeqGAN: Sequence Generative Adversarial Nets with Policy Gradient.

The author-reported oracle NLL on synthetic data is 8.736, against 9.038 for MLE 1609.05473 tab. 1. On real text, SeqGAN is above MLE on each metric the paper reports. On Chinese poems the human score is 0.5356 against 0.4165, and BLEU-2 is 0.7389 against 0.6670 1609.05473 tab. 2. On Obama speeches, BLEU-4 is 0.427 against 0.416 1609.05473 tab. 3. On folk tunes, the BLEU-4 is 0.9406 against 0.9210 1609.05473 tab. 4. On the music task the text says SeqGAN is better on both metrics, so lower MSE is read as better: 20.62 against 22.38 1609.05473 tab. 4. MLE pretraining needs about 150 epochs to converge. A 20-epoch pretraining run makes adversarial training slow and unstable 1609.05473 §Results app. More Ablation Study fig. 4.

The reward has no baseline subtracted. The paper does not discuss the variance that results 1609.05473 §Pitfalls. Each generator step at length 20 with 16 roll-outs runs 304 partial roll-outs and 320 discriminator passes per batch of 64 sequences, in the official code 1609.05473 §Pitfalls. The code is not licensed, and its update order and roll-out copy differ from the paper's Algorithm 1 1609.05473 §Paper vs code. The card cites it by path only.

Hierarchical text generator guided by discriminator features (LeakGAN)

LeakGAN targets long text, where one reward at the end of a sentence gives too little signal. The discriminator is a CNN that reads the sentence so far and exposes its intermediate features. A Manager LSTM maps those features to a unit-length goal at each step. A Worker LSTM picks each next word under a goal embedding built from the last c goals 1709.08624 §Methodology eq. 2 eq. 4. The paper sets c to 4 and the goal size to 16 [arxiv:1709.08624 §Experiment] app. Experiment Settings. The Manager learns from the change in the features over the next c steps, weighted by a Monte Carlo estimate of the expected reward 1709.08624 eq. 7. The Worker learns from an intrinsic reward, the mean cosine between recent feature changes and the goals 1709.08624 eq. 8 eq. 9. Before adversarial training, the paper pretrains the discriminator and generator by maximum likelihood [arxiv:1709.08624 §Training of G] alg. 1. Card: Long Text Generation via Adversarial Training with Leaked Information. The authors' code differs from the paper in several settings, so the card lists the differences.

On synthetic data, the reported negative log-likelihood at length 20 is 7.038, against 8.736 for SeqGAN and 8.247 for RankGAN. At length 40 it is 7.191, against 10.310 and 9.958 1709.08624 tab. 1. On WMT News, BLEU-2 is 0.956, against 0.859 for SeqGAN 1709.08624 tab. 2. On COCO captions, BLEU-5 is 0.686, against 0.544 for RankGAN 1709.08624 tab. 3. In a Turing-style test on WMT News, 55.4% of LeakGAN sentences were judged real, against 23.6% for SeqGAN and 65.1% for real text 1709.08624 tab. 5. The paper reports p < 10⁻⁶ against the second-best method in its main tables 1709.08624 §Evaluation Metrics.

Minimax convex potentials for transport maps

The optimal-transport card learns the quadratic-cost transport map from samples. Two input convex networks, f and g, play a minimax game, and the map is the gradient of g 1908.10962 §3, alg. 1. The page includes it because the paper is tagged as a GAN and compares against a GAN-based transport method, W2GAN 1908.10962 §4.1. Its inner player scores transported points as a convex potential, not as real or fake 1908.10962 §3. Card: Optimal transport mapping via input convex neural networks.

The relative error of the transported mean, scaled by 100 and squared, is 0.02, 0.07 and 0.04 for shifts of 1, 5 and 10 in 784 dimensions [arxiv:1908.10962 tab. 1]. On 2-D toy data, the paper argues that its map does not leave points between separated target modes, which continuous GAN-style generators do 1908.10962 §4.1, §4.3, fig. 3. The bound on the map error needs f to be strongly convex, and the paper does not say how to enforce that 1908.10962 §3.2. Training differentiates through the gradient of g, which needs second-order autodiff 1908.10962 app. C.1.

Conditional GAN with trained encoders for attribute editing (IcGAN)

IcGAN first trains a conditional GAN, then fits two encoders that invert it 1611.06355 §4. The GAN follows DCGAN and uses a matching-aware discriminator, which sees image and label pairs 1611.06355 §4.3. After training, E_z maps a generated image back to its noise vector, and E_y maps a real image to its attribute vector 1611.06355 eq. 3 eq. 4. An edit keeps the noise vector, which the paper links to pose and lighting, and replaces the attributes 1611.06355 §5.4. Of the three ways to couple the two encoders, two independent networks (IND) gives the lowest reconstruction loss 1611.06355 §4.1 §5.3. Card: Invertible Conditional GANs for image editing.

Author-reported reconstruction loss L_e on generated CelebA images is 0.429 for IND, 0.432 for the variant in which E_z also reads the attribute estimate, and 0.500 for a shared encoder 1611.06355 §5.3. Edit quality is reported only qualitatively, apart from attribute-classifier accuracy of 74% to 90% across GAN configurations 1611.06355 tab. 3 §5.2. That classifier is described only by its architecture. The authors placed the attribute input at the generator input and at the first discriminator layer [arxiv:1611.06355 §5.2]. The paper says the encoders use the GAN's optimiser settings, but the released code uses others, so reproduce from the card's listed values 1611.06355 §4.3.

Pretrained GAN reused for attribute editing (StyleFlow)

StyleFlow keeps a pretrained StyleGAN2 frozen. It trains a conditional continuous normalizing flow between the StyleGAN W latent and a standard normal, with 17 attribute values as the condition [arxiv:2008.02401 §5.1, §5.2]. Sampling runs the flow in reverse with the target attributes 2008.02401 §6.1. Editing runs the flow forward under the source attributes, reverses it under the target attributes, and writes the new vector into only the W+ rows that the edit covers. The other rows keep the source values 2008.02401 §6.2. Training maximises a likelihood, not an adversarial objective [arxiv:2008.02401 §5.3, eq. 8]. Card: StyleFlow.

On sequential edits of generated faces, the FID is 53.15, against 64.69 for GANSpace and 67.08 for InterFaceGAN 2008.02401 tab. 1. On real images, the paper says StyleRig scores better on the projection itself 2008.02401 §7.7. Training uses 10,000 samples, 10 epochs and a batch size of 5 2008.02401 §5.1, §5.3. The attribute labels come from an external face-attribute service, so reproduction needs equivalent labels 2008.02401 §5.1. The authors' code is marked non-commercial, so the card's notes apply to any use of it 2008.02401.

Text-only domain shift of a StyleGAN with a directional CLIP loss (StyleGAN-NADA)

StyleGAN-NADA moves a pretrained StyleGAN2 to a new domain from a source text and a target text, with no images from the target domain [arxiv:2108.00946 §1, §4]. Two copies of the generator share one frozen mapping network. The frozen copy gives a source image for each latent code, and the trained copy is updated so that the CLIP-space change between the two images points along the change from the source text to the target text 2108.00946 §4.1. The loss is one minus the cosine between those two directions [arxiv:2108.00946 eq. 2]. The authors reject a global CLIP loss on the trained generator, which they report fails and leads to adversarial pixel patterns and mode collapse 2108.00946 §4.1 app. B fig. 15. For larger shape changes, each iteration updates only the k layers that change most under a short latent optimisation 2108.00946 §4.2. The paper gives k as all layers for texture changes, two thirds of the layers for small shape changes and three for animal changes 2108.00946 app. I tab. 3. A latent mapper is added for large shape changes, where the adapted generator can miss the target 2108.00946 §4.3. Card: StyleGAN-NADA: CLIP-Guided Domain Adaptation of Image Generators.

The FID rows in the card's table are for few-shot methods started from the zero-shot generator, not for the zero-shot method alone. For dog to cat with 10 target images, starting from the zero-shot generator moves Ojha et al. from 45.13 to 41.22, MineGAN from 79.31 to 51.81 and TGAN from 87.11 to 54.49 2108.00946 tab. 2. With 100 images for Ojha et al., the FID goes from 27.54 to 16.17 2108.00946 tab. 2. A short discriminator-only catch-up phase gives the most consistent gains across the three ways tried 2108.00946 tab. 1. In a 2-alternative user study of 1,200 responses from 218 users, the zero-shot cat generator is reported as preferred over most 5- and 10-shot baselines on quality. The text gives no percentages, so read the per-method values from the figure 2108.00946 §5.1 fig. 12. Texture changes take about 300 iterations, roughly 3 minutes on one V100. Animal changes take 2000 iterations plus a latent mapper, about 6 hours in total 2108.00946 app. I.

Text-guided latent optimisation with a frozen VQGAN decoder (VQGAN+CLIP)

VQGAN+CLIP trains no network. It optimises the latent grid of a pretrained VQGAN decoder with Adam, so that CLIP scores the decoded picture closer to the prompt 2204.08583 §2. Each step averages a squared spherical distance over many augmented crops 2204.08583 §2, §2.3. Editing differs from generation only in the starting latent, which is an encoded input image 2204.08583 §2. The paper's settings are a learning rate of 0.15 and 400 iterations 2204.08583 §2. Card: VQGAN-CLIP.

Human raters gave a mean of 4.6 out of 5 on 12 prompts, against 3.3 for the best GLIDE baseline. The baselines were best-of-five and this method was best-of-one 2204.08583 tab. 1, §3.2. One image takes 188 s on a V100 and training costs zero GPU-hours 2204.08583 tab. 2. The paper reports a Colab run under three minutes that needs under 11 GB of GPU memory [arxiv:2204.08583 §6.1].

Latent GAN with a learned seed for point-cloud completion (RL-GAN-Net)

RL-GAN-Net completes a partial 3-D point cloud in a latent space. An autoencoder maps each 2048-point shape to 128 numbers. A latent GAN is trained on those codes, not on raw points, with the WGAN-GP loss 1904.12304 §3.1 §3.2 app.. The GAN's seed is one-dimensional. A DDPG agent reads the code of a partial scan and picks the seed, and the generator's output is decoded into a completed shape 1904.12304 §3.3 alg. 1. The agent's reward is a weighted sum of three negated losses: a Chamfer term, a term that matches the scan's code, and a discriminator term, with weights 100, 10 and 0.01 1904.12304 eq. 5. Each episode is a single action [arxiv:1904.12304 app.]. Card: RL-GAN-Net: A Reinforcement Learning Agent Controlled GAN Network for Real-Time Point Cloud Shape Completion.

The paper reports about 1 ms per completion. The optimisation baseline is quoted at 324 s for a batch of 50 shapes 1904.12304 §4.1 §5. That figure is taken from the baseline's own paper and was not measured on this hardware. With 70% of points missing, PointNet classification reaches 82.5% on the vanilla completions and 83.8% on the hybrid, against 50.2% on the raw partial input 1904.12304 app. §5. The hybrid chooses, per input, between the GAN output and the autoencoder output using the discriminator. It is described only in text 1904.12304 §3.4. The official code departs from the paper in its reward and its Chamfer definition, and it stores every transition as terminal, so the discount factor has no effect 1904.12304. The paper's weights were chosen for its own loss scales.

One tabular GAN per failure type, pooled into the training set

The method oversamples rare machine failures in tabular sensor data. The failure rows are split by failure type, and one WGAN-GP generator is trained per type, each with its own Optuna study 2607.19153 §3.2.5 §3.3.2. The generators come from the tabgan library and use a gradient penalty, mode-specific normalisation for continuous columns and Gumbel-Softmax for categorical ones 2607.19153 §3.3.1. Each generator's synthetic failures are drawn in proportion to its group size and pooled with the real rows. A separate classifier, LightGBM by default, keeps only the synthetic rows it labels as real with high confidence 2607.19153 §3.2.5 §3.3.1. Card: Breaking the Homogeneity Assumption: Specialized Multi-Generator Adversarial Learning for Rare Failure Detection in Predictive Maintenance.

On AI4I 2020 with a Random Forest, the author-reported PR-AUC is 0.8574. That is above 0.835 for cost-sensitive learning and 0.8245 for one GAN over all failures 2607.19153 tab. 4 §4.1. Recall rises from 0.5882 to 0.8824, at a precision of 0.5696 tab. 4. The F1 is 0.6923, below the baseline's 0.740 tab. 4. The SVM baseline's recall of 0.9412 is higher than either multi-GAN row tab. 4. The numbers come from single runs, with no variance or significance test tab. 4. Compute is the main cost: about 14 hours for the multi-GAN and 11 hours 20 minutes for the single GAN, against minutes for the classical methods 2607.19153 §5.1.

Recurrent GAN with feature matching over continuous music events (C-RNN-GAN)

C-RNN-GAN models music as a sequence of continuous events instead of symbols from a fixed menu. Each event has four real values: tone length, frequency, intensity and the time since the previous tone 1611.09904 §2.1. The generator is a two-layer LSTM with 350 units per layer, and it reads a uniform random vector at every step 1611.09904 §3. The discriminator is a two-layer bidirectional LSTM that gives one decision for the whole sequence 1611.09904 §3. Before the adversarial phase, the generator is trained for six epochs to predict the next event. In that phase, the generator's loss is replaced by feature matching on the discriminator's last representation before its output layer, and D is paused while its loss is below 70% of G's 1611.09904 §3. Card: C-RNN-GAN: Continuous recurrent neural networks with adversarial training.

The paper has no numeric table. Its comparisons are curves over epochs, which show more unique tones and a wider tone span than the next-event baseline, and a higher polyphony score for the three-tones-per-cell variant fig. 3a–c. The authors did no formal listening study 1611.09904 §4.1. The code differs from the paper in several settings, including an extra rule that also pauses G and a halved discriminator learning rate, and it has no license file 1611.09904 §Paper vs code.

Residual GAN with a cyclic data-consistency loss for MRI (RefineGAN)

RefineGAN reconstructs an MRI image from under-sampled k-space in one forward pass 1709.00753 §I. The generator is a residual convolutional encoder-decoder. It predicts a correction to the zero-filled image, and the discriminator reuses its encoder path 1709.00753 §III-E. A cyclic loss sits beside the adversarial term. Under-sampling the generator's output again should reproduce the measurements, and a fully sampled image that is under-sampled and reconstructed should come back close to itself 1709.00753 §III-D eq. 6. The adversarial term is weighted 1, the measurement term 1 and the image term 10 1709.00753 §III-F eq. 7. A second generator with the same architecture is chained after the first, and the chain is trained jointly 1709.00753 §III-G. Card: Compressed Sensing MRI Reconstruction using a Generative Adversarial Network with a Cyclic Loss.

The paper reports 0.106 s for brain and 0.112 s for chest on 256×256 images, and 8.57 s and 9.37 s for CSCMRI 1709.00753 tab. 1. The introduction says under 100 ms, which its own RefineGAN row does not meet §I tab. 1. PSNR is shown only as plots, and the text says it passes 42 dB at 40% sampling 1709.00753 §I fig. 8. The paper states that the cyclic loss adds about 1 dB over SingleGAN 1709.00753 §IV-A. Its code has no license and is for research use only, so implement from the paper.

Blind kernel estimation with a linear GAN trained on one image (KernelGAN)

KernelGAN estimates the downscaling kernel of one low-resolution image. It trains a small GAN on that image alone at test time, with no training data 1909.06581 §2. The generator is a deep linear network: five hidden 64-channel layers with no activations, kernel sizes 7, 5, 3, 1 and 1, and a 1×1 output layer with stride 2 1909.06581 §4.1 fig. 4. The kernel is the product of its filters 1909.06581 §4.2. The discriminator is a patch network that scores each local patch 1909.06581 §3. A regularizer keeps the kernel's sum near one, sparse, bounded and centred eq. 3. The estimated kernel is then passed to a non-blind method, ZSSR or SRMD 1909.06581 §5.1. Card: Blind Super-Resolution Kernel Estimation using an Internal-GAN.

Author-reported on 100 DIV2KRK images, KernelGAN with ZSSR reaches 30.363 dB and 0.8669 SSIM at ×2, against 29.368 dB for the earlier estimator with ZSSR, and 26.810 dB at ×4, against 26.080 dB 1909.06581 tab. 1. Each image takes 61 s on a V100 1909.06581 §5.3. Equation 2 as printed gives D the opposite objective to the one the text describes, and the code follows the text 1909.06581 §Pitfalls. The code is for academic and non-commercial use only, so implement from the paper.

Class-conditional GAN for source-free domain adaptation (3C-GAN)

3C-GAN adapts a source-trained classifier to unlabeled target images without using source data 2502.19316 §1 §3. It has three networks. The classifier C supplies the reference weights and is also the model being adapted. A class-conditional generator G(y, z) makes samples for a chosen label. A discriminator D separates real target images from generated ones, with no pairs 2502.19316 §3.1 fig. 2. G is also trained so that C labels its samples as the conditioning label, through a cross-entropy term 2502.19316 eq. 3 eq. 4. C is trained on three terms: cross-entropy on generated samples, a squared-distance anchor to the source weights 2502.19316 eq. 5 eq. 6, and a clustering regularizer on target images 2502.19316 §3.3 eq. 9. Card: Model Adaptation: Unsupervised Domain Adaptation without Source Data.

Author-reported, the paper gives 99.4% on SVHN to MNIST against 76.4% for the source-only model 2502.19316 tab. 1. On the Office-31 average with ResNet50 it gives 89.6% against 76.1% tab. 2, and 81.6% on VisDA17 synthetic to real with ResNet101 tab. 3. Most compared methods use source data during adaptation 2502.19316 §4.2. The text says the gain on MNIST to MNIST-M is around 40%, while the table gives 54.2% rising to 98.5%, about 44 points §4.2 tab. 1. Removing the classifier's cross-entropy term on generated samples stops convergence 2502.19316 §4.4 tab. 4. The semantic weight λ_s is not stated, and no official code was found.

Adversarial distillation of a diffusion teacher (ADD)

Adversarial Diffusion Distillation trains a few-step student from a pretrained diffusion model. The student starts from the teacher's U-Net weights 2311.17042 §3. A hinge-loss discriminator judges the student's samples on frozen DINOv2 features through lightweight heads, with an R1 penalty of 1e-5 2311.17042 §3.2, eq. 2-3. A squared-L2 score-distillation term, weighted by a timestep function, keeps the student close to the teacher, and the final model uses a variant of it 2311.17042 §3.3, eq. 4. The adversarial weight is 2.5 2311.17042 §4. The point of the method is sampling in 1 to 4 steps without classifier-free guidance 2311.17042 §1. Card: Adversarial Diffusion Distillation.

On COCO at 512 px, the ablation reports FID-5k and CLIP score. Adversarial loss alone gives 20.8 and 0.315, the default gives 20.6 and 0.319, and score-distillation without the adversarial term gives 22.3 and 0.325. Random initialisation gives FID 293.6 2311.17042 tab. 1. The larger 860M student (ADD-M) reaches FID 19.7 and CLIP 0.326 in one step 2311.17042 tab. 2. The paper also reports ELO-style preference from figure data, with the one-step large student at 1009 and the four-step at 1107 2311.17042 §4.2. The student is less diverse than its teacher 2311.17042 §4.3. The paper does not state the teacher's timestep distribution or the student's timesteps 2 and 3, and it gives no optimiser, learning rate or batch size. The released code has no training script, and the SDXL-Turbo weights it uses are non-commercial.

Point-cloud rendering with a GAN refinement network (SynSin)

SynSin renders a new view of a scene from one RGB image and a relative camera pose. A feature network gives each pixel a 64-number feature, and a depth network gives each pixel a depth 1912.08804 §3.1. A renderer lifts these into a point cloud, moves it to the new camera and splats each point as a soft disc, keeping the K nearest points per pixel 1912.08804 §3.2 eq. 1 eq. 2. Pixels that no point reaches are holes. A refinement network with eight residual blocks fills them, and two discriminators at different scales judge its output with a feature-matching term 1912.08804 §3.3. The generator loss is the GAN term with weight 1, plus L1 with weight 1 and a perceptual term with weight 10 1912.08804 §3.4. Training uses image pairs and their relative poses, with no depth or 3D labels 1912.08804 §1 §3.4. Card: SynSin: End-to-end View Synthesis from a Single Image.

Author-reported PSNR on RealEstate10K is 22.31 with one input view. The Vox baseline (its "Vox w/ ours") scores 21.88 in the same table, and 3DView scores 21.88 in the system comparison. StereoMag reaches 25.34, but it takes two input views at test time 1912.08804 tab. 1 tab. 2 §4.2. In a five-rater study on 100 animated trajectories, SynSin was preferred over Vox w/ ours in 68.7% of either-or votes 1912.08804 tab. 4. On Matterport3D the full model scores 20.91 PSNR 1912.08804 tab. 1. The ablations in the card change the depth supervision, the input features, the point count and the fine-tuning setup, but none removes the GAN term. So the card cannot say how much the adversarial loss adds 1912.08804 tab. 1.

The paper's learning rates conflict with each other and with the code. The text gives 0.0001 for the generator and 0.01 for the discriminator, and the appendix gives 0.001 for its own model 1912.08804 §3.4 app. C. The code's default refinement network is a UNet, while the released scripts use the ResNet variant the paper describes 1912.08804 §Paper vs code. The official code mixes licences, including non-commercial CC BY-NC-SA parts, so implement from the paper and check the components you reuse.

Motion-conditioned video GAN on a StyleGAN2 backbone (StyleGAN-V)

StyleGAN-V keeps StyleGAN2 for image synthesis and adds time. A content code, passed through the StyleGAN2 mapping network, gives each clip its look. A grid of motion noise is passed through two padding-less one-dimensional convolutions, and an acyclic sinusoidal embedding turns the result into a motion code for any real-valued time 2112.14683 §3.1 app. B. The motion code is tiled into the constant input of the synthesis network, so frames are made independently and none feeds the next 2112.14683 §3.1. The discriminator takes k frames of a clip, extracts features from each with a StyleGAN2-style backbone, joins them at 16 by 16, and is conditioned on the time gaps between frames 2112.14683 §3.2 fig. 6. Training frames are sampled with at most 32 frames of total span 2112.14683 §3.3. The author-reported FVD16 at 256 squared is 47.41 on FaceForensics, 79.52 on SkyTimelapse and 1431.0 on UCF101, the lowest in the authors' table on each of those sets. VideoGPT is lowest on RainbowJelly 2112.14683 tab. 1. On FaceForensics the authors report FID 9.44 against 8.42 for StyleGAN2 2112.14683 §4.3. The official code has no LICENSE file, and it differs from the paper in several places, so the paper's formulas do not give its behaviour 2112.14683 §Paper vs code. Card: StyleGAN-V: A Continuous Video Generator with the Price, Image Quality and Perks of StyleGAN2.

When to use which

Implementation sketch

Common core of the adversarial variants, in the form the cards share. The variants differ in the extra terms, in how D scores its input, and in the loss form.

# Adversarial training step, common core  [arxiv:1612.05362 §2.1]
# G is conditioned on an input x, not on noise
# Variant terms are passed in as `extra`

def gan_step(x, y, w_adv, extra):
    # Discriminator: real target -> 1, generated -> 0  [eq. 3, eq. 4]
    fake = G(x)
    update(D, bce(D(y), 1) + bce(D(detach(fake)), 0))

    # Generator: fool D, plus the variant's extra terms  [eq. 5, eq. 7]
    loss = w_adv * bce(D(fake), 1) + extra(x, fake, y)
    update(G, loss)

Variant notes. For DTN, D has three classes, G is g(f(x)), f stays frozen, and no paired y is used. The extra terms are feature constancy on source samples and identity on target samples 1611.02200 eq. 3–6. For the MR-to-CT variant, extra is an L2 term on the paired CT plus a 3D gradient-difference term 1612.05362 eq. 5, eq. 6. Its discriminator is a single-output CNN with binary cross-entropy [arxiv:1612.05362 eq. 3, eq. 4]. For the motion variant, the critics use a least-squares form: real windows are pushed to 1 and generated ones to 0, each with a factor of one half 2102.04942 §3.6.2. The fashion variant runs this step once per stage, with the conditions given to D as well 1710.07346 §3.4. Learning rate, batch size and iteration counts are not stated in either DTN or fashion text, so treat them as tuning parameters 1611.02200 §4 1710.07346 §3.4.

For the aerial translator, the step runs in both directions. The reconstruction term is the cycle loss on each round trip, which the paper leaves unweighted and the code weights at 10. The code also adds an identity term that the paper does not describe 1905.03198 eq. 8 §4.3. For SD-GAN, extra is the contrastive term on the discriminator features of the two branches, added to the AttnGAN losses with no weight given 1904.01480 §4.1. For RL-GAN-Net, the GAN is trained first with a WGAN-GP critic on latent codes. The agent is trained afterwards against a frozen generator, so its reward does not update G 1904.12304 §3.2 §3.3.

For 3C-GAN, G takes a label as well as noise, and D scores real target images against generated ones without pairs. C is trained on G's samples with their conditioning labels, so it also acts as the semantic judge for G 2502.19316 §3.1 alg. 1. For RefineGAN, extra is the cyclic loss, and D reads the output of the chained generator 1709.00753 eq. 6 §III-G. For KernelGAN, both networks see crops of one image. D scores patches with an L1 loss in least-squares form, and the kernel regularizer is added to G's loss 1909.06581 §2 §4.2. For C-RNN-GAN, D reads a whole sequence and gives one decision, and feature matching replaces G's adversarial loss in the adversarial phase 1611.09904 §3. For StyleGAN-V, D scores a whole clip, so its logit and its R1 penalty are per clip. The loss is the non-saturating logistic form 2112.14683 app. B. Its generator is a StyleGAN2 network with the motion code as its constant input [arxiv:2112.14683 §3.1].

LeakGAN does not fit this core. Its generator is a word-level policy, and its Manager and Worker take policy-gradient updates from feature changes and an intrinsic reward, not one adversarial term 1709.08624 eq. 7 eq. 8. StyleGAN-NADA does not fit it either. It has no discriminator, and its update is one minus the cosine between a CLIP image direction and a CLIP text direction 2108.00946 eq. 2. GANFIT and DragGAN do not fit it, because neither trains a discriminator. GANFIT optimises a latent through a face recognition loss, and DragGAN through a feature-matching loss on handle points 1902.05978 §3.2 2305.10973 eq. 1.

Contradictions

None known between these cards on the same claim. The StyleGAN-NADA card reports that a global CLIP loss on the trained generator fails and collapses 2108.00946 §4.1 fig. 15. VQGAN+CLIP optimises a latent with the same global CLIP loss and reports strong human ratings 2204.08583 §2 2204.08583 tab. 1. The two optimise different things, generator weights in one and a latent of a frozen decoder in the other, and the cards do not test the global loss on the other's setting, so the results do not conflict. Four further points keep the cards from being compared directly. The DTN card says a direct binary GAN with feature constancy on raw pixels "does not produce desirable results" and scores 13.71% on its digit task 1611.02200 §3, tab. 1. The MR-to-CT card uses a paired generator with L2 and reports PSNR gains from the adversarial term, on a different dataset and metric 1612.05362 §4.1. The motion card reports that its adversarial term adds little to L2Q 2102.04942 tab. 2. The aerial card's gain is a segmentation accuracy, so it does not bear on the PSNR gain in the MR-to-CT card 1905.03198 tab. 3 1612.05362 §4.1. Two cards compare an adversarial text generator with maximum likelihood, on different corpora. SeqGAN is above MLE on BLEU-4 for Obama speeches, 0.427 against 0.416 1609.05473 tab. 3. Texygen's own runs on COCO captions put SeqGAN below MLE, 0.294 against 0.305 1802.01886 tab. 2. The corpora and implementations differ, so the cards do not show which sign holds in general. All of these numbers are author-reported. The LeakGAN text results come from the authors' own runs against SeqGAN and RankGAN, and the wiki holds no independent reproduction of them 1709.08624 §Experiment. The underwater card's ablation has no run without the adversarial term, so it adds no evidence on whether that term helps 2111.11843 tab. 6. The tabular card disagrees with itself. Its text calls the multi-GAN F1 the highest of all conditions and also says it is slightly below the baseline. Table 4 gives 0.6923 for the multi-GAN and 0.740 for the baseline, so the table is the value to use 2607.19153 §4.1 tab. 4. Its baseline PR-AUC is 0.835 in the table and 0.836 in a figure tab. 4 fig. 11.

The two inversion cards report incompatible numbers. PTI gives LPIPS, MSE, MS-SSIM and identity similarity on CelebA-HQ 2106.05744 tab. 1. Image2StyleGAN++ gives PSNR and, for inpainting, SSIM, MSE and PSNR on its own images 1911.11544 tab. 1. Their reconstruction gains cannot be ranked against each other. PTI also reports that a mean or random code used as a fixed pivot gives weaker edits 2106.05744 §4.4. Image2StyleGAN++ starts its optimisation from a mean face code 1911.11544 §7.1. The two uses differ, so the cards do not conflict. The augmentation benchmark disagrees with itself. Its conclusion says CT-GAN is not a top performer, but its table lists CT-GAN among the best pairs on five of 23 binary sets 2304.02858 §7 tab. 8.

The video card reports FVD values on its own protocol, with every baseline retrained by its authors. Its FVD numbers therefore cannot be set against FVD values from other papers 2112.14683 §4 app. C. The FVD card says that identical sets have a nonzero distance and that comparisons need equal sample counts 1812.01717 §4.2.

Fitting a 3D face through a GAN texture prior (GANFIT)

GANFIT recovers shape and skin of a 3D face from one photo by gradient descent. The shape is a linear 3D morphable model, with identity and expression bases. The skin is a generator trained on about 10,000 UV textures. The texture equation optimises a 512-value input latent; whether the generator's own weights are also updated is not stated 1902.05978 §3.1 §4.1 eq. 2 eq. 5. A differentiable renderer produces the image, and the loss mixes a cosine distance on face-recognition embeddings, intermediate face-network features, a pixel L1 term and a landmark term 1902.05978 §3.2 §3.3 eq. 8–11. Fitting runs in two stages: a landmark-only alignment, then a joint stage over all parameters 1902.05978 §3.4. Card: GANFIT: Generative Adversarial Network Fitting for High Fidelity 3D Face Reconstruction.

On the MICC Florence benchmark, the author-reported mean shape error in each of three settings is 0.94 to 0.95 mm, against 1.48 to 1.50 for the best baseline in each setting 1902.05978 tab. 1. The text gives a 36% drop against the other two methods, and the Genova et al. row matches that figure §4.3. The identity results on LFW are given only as cosine similarities app. A. The ablation of the identity, content and pixel terms is qualitative §4.4. The text says the identity latent is averaged across frames, but the equation sums them §3.4 eq. 13. The released repository has the evaluation scripts only, not the fitting code [arxiv:1902.05978 §Paper vs code], so the hyperparameters come from the paper alone.

Point-based editing of a StyleGAN2 image by motion supervision (DragGAN)

DragGAN moves chosen handle points of a pretrained StyleGAN2 image to target points. Each iteration takes one optimiser step on the latent code. The loss pushes the features near each handle toward its target, using the features after the sixth generator block 2305.10973 §3.2 eq. 1. Each handle is then re-located by nearest-neighbour search in the same features 2305.10973 §3.3 eq. 2. No optical-flow or tracking network is used. The loop stops when every handle is within a tolerance of its target §3.4. Card: Drag Your GAN: Interactive Point-based Manipulation on the Generative Image Manifold.

On a face-landmark benchmark with one point, the author-reported mean landmark distance is 2.44, against 11.64 for UserControllableLT and 13.43 for the same method with RAFT tracking 2305.10973 tab. 1. The edit takes about 2.0 s per run tab. 1. The paper names an RTX 3090 in its introduction but does not say that Table 1 was timed on it §1. For real images the paper first inverts with PTI, and results depend on that inversion 2305.10973 §4.1 fig. 5. The paper reports 30 to 200 iterations in typical use, but the step length is not fixed, so that count is an observation and not a stopping rule §3.1. The distance unit is not stated tab. 1. The repository's licence is the NVIDIA Source Code License for StyleGAN3, which allows research use only, so implement from the paper 2305.10973 §Paper vs code.

Papers

• Unsupervised Cross-Domain Image Generation: DTN, unpaired translation through a frozen feature function, with a three-class discriminator and identity and constancy terms.
• Medical Image Synthesis with Context-Aware Generative Adversarial Networks: paired 3D conditional GAN for MR-to-CT synthesis, with an L2 and gradient-difference loss and Auto-Context stacking.
• Be Your Own Prada: Fashion Synthesis with Structural Coherence: two conditional GANs, one for a segmentation map and one for per-label textures, for text-guided outfit changes.
• Robust Motion In-betweening: LSTM generator with two least-squares critics and additive time-to-arrival and target-noise embeddings for transition generation.
• Optimal transport mapping via input convex neural networks: transport map as the gradient of a convex network, learned by a minimax game against a second convex network.
• StyleFlow: conditional flow on the StyleGAN latent for attribute sampling and editing, with no adversarial training.
• StyleGAN-NADA: CLIP-Guided Domain Adaptation of Image Generators: text-only domain shift of a StyleGAN2 generator with a directional CLIP loss and adaptive layer freezing.
• VQGAN-CLIP: training-free text-guided generation and editing by optimising a frozen VQGAN latent against CLIP.
• SeqGAN: Sequence Generative Adversarial Nets with Policy Gradient: a text generator as a policy, with a discriminator that scores finished sequences and roll-out estimates of token values.
• StarGAN v2: Diverse Image Synthesis for Multiple Domains: one generator for many domains, with style codes from noise or a reference image, injected by AdaIN, and diversity and cycle terms.
• Texygen: A Benchmarking Platform for Text Generation Models: a benchmark harness with seven text generators and five metrics. It adds no method of its own.
• Long Text Generation via Adversarial Training with Leaked Information: LeakGAN, a text GAN whose Manager reads discriminator features and steers a Worker's word choice.
• Semantics Disentangling for Text-to-Image Generation: SD-GAN, a stacked text-to-image GAN with a Siamese contrastive term and semantic-conditioned batch norm.
• RL-GAN-Net: a latent GAN on autoencoder codes, with a DDPG agent that picks the seed for point-cloud completion.
• Unsupervised Domain Adaptation using Generative Adversarial Networks for Semantic Segmentation of Aerial Images: cycle-consistent restyling of labelled source images, then fine-tuning a segmenter without target labels.
• WaveFake: a dataset of generated speech, some from GAN-based vocoders, for audio deepfake detection. The GAN is not the method.
• A Survey on Bridging EEG Signals and Generative AI: a literature survey in which GANs are one image decoder among several. It reports no new results.
• The History Began from AlexNet: A Comprehensive Survey on Deep Learning Approaches: a deep-learning survey with one section on GANs. It adds no method and no code.
• Invertible Conditional GANs for image editing: IcGAN, a conditional GAN plus two encoders that map a real image to a noise vector and an attribute vector, for attribute edits.
• U-shape Transformer for Underwater Image Enhancement: a GAN-trained U-shaped generator with channel-wise and spatial transformer blocks and LAB and LCH colour losses, with the LSUI dataset.
• Breaking the Homogeneity Assumption: Specialized Multi-Generator Adversarial Learning for Rare Failure Detection in Predictive Maintenance: one WGAN-GP tabular generator per failure type, pooled into the training set to oversample rare failures.
• C-RNN-GAN: Continuous recurrent neural networks with adversarial training: GAN over continuous music events, with an LSTM generator, a bidirectional LSTM discriminator and feature matching.
• Compressed Sensing MRI Reconstruction using a Generative Adversarial Network with a Cyclic Loss: residual GAN with a cyclic data-consistency loss and a chained second generator, for MRI reconstruction.
• Blind Super-Resolution Kernel Estimation using an Internal-GAN: KernelGAN, a linear generator trained on one image to estimate its blur kernel for super-resolution.
• Image2StyleGAN++: How to Edit the Embedded Images?: embedding a photo into StyleGAN's W+ space and then into noise maps, with masks for local edits and inpainting.
• Pivotal Tuning for Latent-based Editing of Real Images: PTI, an inversion to the W space followed by a short generator fine-tune with a locality penalty, for editing real faces.
• Ensemble and augmentation benchmark for imbalanced data: a benchmark of nine augmenters, CT-GAN among them, combined with ensemble models on imbalanced data.
• Model Adaptation: Unsupervised Domain Adaptation without Source Data: 3C-GAN, a class-conditional generator and classifier for source-free domain adaptation.
• Age Progression/Regression by Conditional Adversarial Autoencoder: a conditional autoencoder with an age-conditioned image discriminator and a prior-matching discriminator on the code, for age progression and regression.
• Adversarial Diffusion Distillation: a few-step student distilled from a diffusion teacher, with a hinge discriminator on frozen features.
• SynSin: End-to-end View Synthesis from a Single Image: single-image view synthesis with a point-cloud renderer, a depth network, and a refinement GAN with two discriminators.
• Social GAN forecasts for robot navigation among people: a Social GAN forecaster feeds extra social cost terms into the Timed Elastic Band local planner.
• Fréchet Video Distance: an evaluation metric for video generators, computed on Kinetics-400 features. It trains no GAN.
• Deep Generative Modelling: A Comparative Review of VAEs, GANs, Normalizing Flows, Energy-Based and Autoregressive Models: a survey that compares five generative families, GANs among them, on cited CIFAR-10 numbers. It adds no method.
• StyleGAN-V: A Continuous Video Generator with the Price, Image Quality and Perks of StyleGAN2: a StyleGAN2-based video GAN with motion codes for any real-valued time and a discriminator on k sampled frames.
• Beyond Statistical Similarity: Rethinking Metrics for Deep Generative Models in Engineering Design: a review of metric groups for generative models in engineering design, with a GAN and a VAE on 2-D data as the running example.
• GANFIT: Generative Adversarial Network Fitting for High Fidelity 3D Face Reconstruction: a 3D morphable model fitted to one photo, with a GAN texture prior and face-recognition losses.
• Drag Your GAN: Interactive Point-based Manipulation on the Generative Image Manifold: point-based editing of a pretrained StyleGAN2 image, with motion supervision on the latent and nearest-neighbour point tracking.
• Pros and Cons of GAN Evaluation Measures: New Developments: a survey of GAN evaluation measures published after 2019, with no new experiments.
• A Review of Graph Neural Networks and Their Applications in Power Systems: a review in which graph GANs are one of several graph generative models, with one quoted synthetic-feeder result.
• Deep Partial Multi-View Learning: CPM-Nets, shared per-sample codes rebuilt into observed views, with an optional per-view GAN (CPM-GAN) for missing entries.
• Emotional Voice Conversion: Theory, Databases and ESD: the ESD emotional speech database, benchmarked with CycleGAN-EVC and VAWGAN-EVC. The GAN is a baseline, not the contribution.