Pith. sign in

REVIEW 3 major objections 4 minor 52 references

Multi-population GAN Training: Analyzing Co-Evolutionary Algorithms

T0 review · 3 major / 4 minor · reviewed 2026-08-06 · deepseek-v4-flash

Pith's one-line read This paper claims that (mu,lambda) full generational replacement consistently yields better GAN sample quality and diversity than (mu+lambda) elitist or tournament schemes.

desk verdict A competent empirical comparison of replacement strategies in coevolutionary GANs; the core (mu,lambda) claim survives compute-matched checks, but the larger-offspring advantage is confounded by an unscaled budget. read the letter →

arxiv 2507.13157 v1 pith:L4STFOJ5 submitted 2025-07-17 cs.NE

classification cs.NE
keywords generativeadversarialnetworkscoevolutionaryalgorithmspopulation-basedtrainingmodecollapseselectionandreplacementsamplediversityMNISTWassersteindistance
verification ladder T0 review T1 audit T2 compute T3 formal

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The reading

This paper tests three ways of managing populations of generators and discriminators during GAN training: full generational replacement $(\mu,\lambda)$, elitist $(\mu+\lambda)$, tournament-based $(\mu+\lambda)$, plus a non-evolutionary multi-generator multi-discriminator baseline. It claims that $(\mu,\lambda)$, which discards the parent population every generation, consistently outperforms the others on both sample quality and diversity across synthetic 2D datasets and MNIST. Elitist strategies tend to converge prematurely and lose diversity, while tournament selection helps but only when paired with full replacement. A static population without evolution works on simple data but collapses on structured or high-dimensional distributions.

What carries the argument

The load-bearing mechanism is the population update rule in Algorithm 1, specifically the replacement step. $(\mu,\lambda)$ empties the parent population each generation and retains the $\mu$ best offspring, forcing broad exploration; $(\mu+\lambda)$ selects the $\mu$ best individuals from the union of parents and offspring, encouraging exploitation through elitism. The paper also varies offspring size $\lambda$ to control the volume of candidate solutions explored per generation.

What would settle it

Run every method under a fixed total number of training epochs rather than a budget that scales with offspring size; if $(\mu,\lambda)$'s advantage vanishes or reverses, the claimed exploration benefit is a budget effect.

Watch

Extended reading notes

Core claim

The central discovery is that the replacement step in coevolutionary GAN training matters more than the choice of selection operator. In Algorithm 1, $(\mu,\lambda)$ keeps only the $\mu$ fittest of the $\lambda$ offspring and drops all parents, while $(\mu+\lambda)$ keeps the $\mu$ fittest from parents plus offspring. Across all tested configurations, the full-replacement scheme achieves the best Wasserstein distance on synthetic data, the best FID on MNIST, the best total variation distance between generated and real label distributions, and perfect mode coverage on the structured gaussian datasets. Elitist $(\mu+\lambda)$ performs worst, and tournament-based $(\mu+\lambda)$ sits in between, suggesting that retaining high-performing individuals too aggressively reduces exploration and leads to mode collapse.

Load-bearing premise

The comparison assumes that a training budget of $250\lambda$ epochs is fair across offspring sizes, so larger $\lambda$ is not merely buying more compute.

Editorial extensions

If this is right

  • Designers of coevolutionary GANs should prefer $(\mu,\lambda)$ full generational replacement over elitist $(\mu+\lambda)$ strategies.
  • Increasing offspring size improves quality and diversity mainly when combined with full replacement, not when elites are retained.
  • Elitism in coevolutionary GAN training tends to cause premature convergence and mode collapse, especially on structured distributions.
  • A static multi-generator multi-discriminator population without evolution is competitive on simple 2D data but fails on structured or high-dimensional data.
  • Tournament selection is a better complement to full replacement than pure elitism, but it does not substitute for discarding parents.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

  • If the compute-budget confound is resolved, the paper's RQ2 conclusion that larger offspring helps may shrink, but the RQ1 conclusion about replacement strategy would likely stand.
  • The full-replacement advantage may act as an implicit regularizer, preventing early elites from dominating the fitness landscape; tracking per-generation loss curves could test this directly.
  • A natural extension is to test whether the same ordering holds on higher-resolution image benchmarks, where mode collapse is harder to detect with FID alone.
Share X Bluesky LinkedIn Reddit HN

Signed reviews

No signed human review yet.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Referee Report

3 major / 4 minor

Summary. The paper presents an empirical comparison of three coevolutionary GAN training schemes — (μ,λ) with full generational replacement, (μ+λ) with elitism, and (μ+λ) with tournament selection — together with a non-evolutionary multi-generator multi-discriminator (MG-MD) baseline. Experiments are run on synthetic 2D Gaussian mixtures (blob, gaussian-2/4/8) and MNIST, with quality and diversity measured by Wasserstein distance, FID, mode coverage, cluster entropy, and TVD. The central claim is that (μ,λ) full-replacement training consistently yields better sample quality and diversity than elitist or tournament-based (μ+λ) variants, particularly with larger offspring sizes, while MG-MD is competitive only on simple datasets.

Significance. If the central claim is accepted, the paper would provide useful practical guidance for designing coevolutionary GAN training, namely that full generational replacement is preferable to elitist retention. The study's strengths include the use of external, established metrics (WD, FID, TVD, mode coverage), 30 independent runs per configuration, and a matched λ=μ comparison in Table 5 where (μ,λ) indeed achieves lower FID than both (μ+λ) variants. However, the broader scope of the claim — especially the advantage of larger offspring sizes — is weakened by a compute-budget confound, and the MG-MD baseline's budget is not specified. These issues are addressable with additional matched-budget experiments or appropriately qualified conclusions.

major comments (3)
  1. [Section 4, RQ3] The training budget is not compute-matched across configurations, so the reported advantage of larger offspring sizes is confounded. With T_B = 250λ, the number of generations is floor(T_B/(n_t·λ)) = 50, and total training epochs equal 250λ for every method; therefore λ=2μ receives twice the epochs of λ=μ. Since the design uses λ∈{μ, 1.5μ, 2μ} for (μ,λ) and λ∈{1, ⌈μ/2⌉, μ} for (μ+λ), every (μ,λ) configuration has at least as large a budget as the largest (μ+λ) configuration. Consequently, the cross-λ comparisons in Figure 2 and Tables 1–4, as well as the RQ2 and Abstract claim that 'increasing offspring size generally improves performance,' are not supported by the current design. The λ=μ comparisons in Table 5 are matched and do support the replacement-strategy advantage, but the larger-offspring conclusion requires either a fixed total-epoch budget across configurations or an explicit matched-budget control.
  2. [Section 4, RQ3] The MG-MD baseline's compute budget is unspecified. The text sets T_B = 250λ only for the coevolutionary variants; for MG-MD there is no equivalent total-epoch or iteration count. Since MG-MD is reported as collapsing on gaussian-4 and underperforming on MNIST (Section 5.1, Table 5), the conclusion that evolutionary dynamics are necessary cannot be separated from a possible smaller compute budget. Please specify the MG-MD training schedule and equalize it with the coevolutionary methods.
  3. [Abstract, Section 5.1] The claim of consistent outperformance is stronger than the reported statistics. Section 5.1 states that for (μ,λ), 'improvements are generally limited and not always statistically significant according to the Wilcoxon rank-sum test.' The abstract's 'consistently outperforms in terms of both sample quality and diversity' should be qualified to the configurations and metrics where the difference is statistically significant.
minor comments (4)
  1. [Section 4] The text says the gaussian datasets comprise 'two, four, and ten 2D Gaussian modes' but the datasets are named gaussian-2, gaussian-4, and gaussian-8; 'ten' should be 'eight' to be consistent.
  2. [Table 2] The table heading reads '(μ,λ)E', which is inconsistent with the notation used elsewhere; it should be '(μ,λ)' or the 'E' should be defined and used consistently.
  3. [Table 6] In the row for μ=5, λ=2μ, the TVD entry contains an extra closing parenthesis: '1.467 (0.206))' should be '1.467 (0.206)'.
  4. [Section 4] The sentence 'Both blob and gaussian datasets consist of 2D vectors within the range [−1, 1] and are divided into ten clusters (i.e., Gaussian distributions)' conflicts with the earlier description of blob as having eight modes; the cluster count should be reconciled.

Circularity Check

0 steps flagged · score 0.0 of 10

No circularity: the paper's conclusions are empirical measurements against external benchmarks, with no derivation reducing to fitted inputs or self-citations.

full rationale

This paper is an empirical benchmark study, not a derivation chain. All central conclusions are supported by measured Wasserstein distance, FID, mode coverage, cluster entropy, and TVD computed against the external data distributions (synthetic Gaussian mixtures and MNIST), rather than by equations whose outputs are equivalent to their inputs. The adoption of hyperparameters from Sedeño et al. [35], which includes coauthor Toutouh, is a prior-configuration choice and is not load-bearing for the replacement-strategy comparison; the same holds for the background citations to the authors' earlier coevolutionary GAN work. The potential compute-budget confound from setting T_B = 250λ is a real experimental-design concern about fairness across λ values, but it is a correctness/validity risk, not a circularity: the (μ,λ) advantage over (μ+λ) at matched λ is a direct empirical observation, and no fitted parameter is renamed as a prediction. No self-definitional, fitted-input, uniqueness-importation, or renamed-known-result pattern is present. Hence no significant circularity: score 0.

Assumptions & free parameters 3 free parameters · 3 assumptions · 0 invented entities

The paper is empirical and introduces no new theoretical entities. Its conclusions rest on hand-chosen hyperparameters and on domain assumptions about metric validity and compute fairness.

free parameters (3)
  • Total training epochs scaling T_B = 250lambda = 250lambda (e.g., 750 for lambda=3, 3500 for lambda=14)
    Chosen so the number of generations is fixed at 50; this makes total compute grow with offspring size and confounds comparisons across lambda.
  • Tournament size tau = not specified
    Algorithm 1 takes tau as a selection parameter but no value is given in Section 4, leaving selection pressure undefined.
  • Number of training epochs per couple n_t = 5
    Set by hand; the paper does not test sensitivity to this value.
assumptions (3)
  • domain assumption The chosen metrics (Wasserstein distance, FID, mode coverage, cluster entropy, TVD) are valid proxies for sample quality and diversity.
    The central claim is based on these metrics; no justification is given for their validity in this setting (Section 4).
  • domain assumption Comparison across offspring sizes is fair even though total training epochs scale with lambda (T_B = 250lambda).
    The paper implicitly assumes that larger lambda improves performance due to exploration rather than extra compute (Section 4).
  • domain assumption The selected datasets and architectures are representative enough to support a general conclusion about coevolutionary GAN training.
    Only three families of 2D synthetic datasets plus MNIST are used; the paper generalizes beyond them in the conclusions (Section 5.3).

how reviews work

0 comments
Cite this review

Pith. "Pith review of Multi-population GAN Training: Analyzing Co-Evolutionary Algorithms." pith.science (2026). https://pith.science/paper/L4STFOJ5

@misc{pith2026250713157,
  author       = {Pith},
  title        = {Pith review of: Multi-population GAN Training: Analyzing Co-Evolutionary Algorithms},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/L4STFOJ5}},
  note         = {Machine review of arXiv:2507.13157}
}
read the original abstract

Generative adversarial networks (GANs) are powerful generative models but remain challenging to train due to pathologies suchas mode collapse and instability. Recent research has explored co-evolutionary approaches, in which populations of generators and discriminators are evolved, as a promising solution. This paper presents an empirical analysis of different coevolutionary GAN training strategies, focusing on the impact of selection and replacement mechanisms. We compare (mu,lambda), (mu+lambda) with elitism, and (mu+lambda) with tournament selection coevolutionary schemes, along with a non-evolutionary population based multi-generator multi-discriminator GAN baseline, across both synthetic low-dimensional datasets (blob and gaussian mixtures) and an image-based benchmark (MNIST). Results show that full generational replacement, i.e., (mu,lambda), consistently outperforms in terms of both sample quality and diversity, particularly when combined with larger offspring sizes. In contrast, elitist approaches tend to converge prematurely and suffer from reduced diversity. These findings highlight the importance of balancing exploration and exploitation dynamics in coevolutionary GAN training and provide guidance for designing more effective population-based generative models.

Figures

Figures reproduced from arXiv: 2507.13157 by the authors.

Figure 1
Figure 1. Representation of the datasets applied in our em [PITH_FULL_IMAGE:figures/full_fig_p003_1.png] view at source ↗
Figure 2
Figure 2. Influence of 𝜇 and 𝜆 size in 𝑊𝐷 on the blob and gaussian datasets. gaussian-4 datasets. As shown in [PITH_FULL_IMAGE:figures/full_fig_p005_2.png] view at source ↗
Figure 3
Figure 3. blob-8 samples produced when using 𝜇=3. (a) (𝜇 + 𝜆)𝐸 (b) (𝜇 + 𝜆)𝑇 (c) (𝜇, 𝜆) (d) MG-MD GAN [PITH_FULL_IMAGE:figures/full_fig_p006_3.png] view at source ↗
Figures from the paper (4 more)
Figure 4
Figure 4. Figure 4: gaussian-2 samples produced when using 𝜇=3. (a) (𝜇 + 𝜆)𝐸 (b) (𝜇 + 𝜆)𝑇 (c) (𝜇, 𝜆) (d) MG-MD GAN [PITH_FULL_IMAGE:figures/full_fig_p006_4.png]
Figure 5
Figure 5. Figure 5: gaussian-4 samples produced when using 𝜇=3. A general trend can be seen in the visual results in figures 3-6 in which (𝜇, 𝜆) GAN produced samples closer to the real dataset with fewer samples out of the modes that define the datasets than the other approaches. Focusing…
Figure 6
Figure 6. Figure 6: gaussian-8 samples produced when using 𝜇=3. modes (see [PITH_FULL_IMAGE:figures/full_fig_p007_6.png]
Figure 7
Figure 7. Figure 7: shows some samples produced by the analyzed ap￾proaches when using population size 3. It can be observed that in general all the approaches produced high quality samples. However, (𝜇 + 𝜆)𝐸 samples set has lower diversity, i.e., high frequency of one digit, than the set…

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

52 extracted references · 48 canonical work pages

  1. [1]

    Mohd Ali, Mehboob Ali, Mubashir Hussain, and Deepika Koundal. 2024. Genera- tive adversarial networks (gans) for medical image processing: Recent advance- ments. Archives of Computational Methods in Engineering (2024), 1–14

  2. [2]

    Luca Ambrogioni, Tue Herlau, and Mathias Bastian. 2021. Automatic evaluation of generative models using density-ratio estimation. Neural Computation 33, 5 (2021), 1119–1140

  3. [3]

    L. M. Antonio and C. A. C. Coello. 2018. Coevolutionary Multi-objective Evo- lutionary Algorithms: A Survey of the State-of-the-Art. IEEE Transactions on Evolutionary Computation (2018), 1–16. doi:10.1109/TEVC.2017.2767023

  4. [4]

    Andrea Arcuri and Xin Yao. 2007. Coevolving programs and unit tests from their specification. In Proceedings of the 22nd IEEE/ACM International Conference on Automated Software Engineering. 397–400

  5. [5]

    Andrea Arcuri and Xin Yao. 2014. Co-evolutionary automatic programming for software development. Information Sciences 259 (2014), 412–432

  6. [6]

    Martin Arjovsky, Soumith Chintala, and Léon Bottou. 2017. Wasserstein genera- tive adversarial networks. In International conference on machine learning . PMLR, 214–223

  7. [7]

    Sanjeev Arora, Andrej Risteski, and Yi Zhang. 2018. Do GANs learn the dis- tribution? some theory and empirics. In International conference on learning representations

  8. [8]

    Robert Axelrod and William D Hamilton. 1981. The evolution of cooperation. science 211, 4489 (1981), 1390–1396

Show all 52 references
  1. [9]

    Jing Bai, Yang Zhang, Zhu Xiao, Fawang Ye, You Li, Mamoun Alazab, and Licheng Jiao. 2022. Immune evolutionary generative adversarial networks for hyperspec- tral image classification. IEEE Transactions on Geoscience and Remote Sensing 60 (2022), 1–14. Multi-population GAN Trai...

  2. [10]

    Marco Baioletti, Carlos Artemio Coello Coello, Gabriele Di Bari, and Valentina Poggioni. 2020. Multi-objective evolutionary GAN. In Proceedings of the 2020 genetic and evolutionary computation conference companion . 1824–1831

  3. [11]

    Earl T Barr, Mark Harman, Phil McMinn, Muzammil Shahbaz, and Shin Yoo. 2014. The oracle problem in software testing: A survey. IEEE transactions on software engineering 41, 5 (2014), 507–525

  4. [12]

    Haripriya Chakraborty, Priyanka Samanta, and Liang Zhao. 2021. Sequential data imputation with evolving generative adversarial networks. In 2021 International Joint Conference on Neural Networks (IJCNN) . IEEE, 1–8

  5. [13]

    Victor Costa, Nuno Lourenço, João Correia, and Penousal Machado. 2019. CO- EGAN: evaluating the coevolution effect in generative adversarial networks. In Proceedings of the genetic and evolutionary computation conference . 374–382

  6. [14]

    Victor Costa, Nuno Lourenço, João Correia, and Penousal Machado. 2020. Neu- roevolution of generative adversarial networks. Deep Neural Evolution: Deep Learning with Evolutionary Computation (2020), 293–322

  7. [15]

    Victor Costa, Nuno Lourenço, João Correia, and Penousal Machado. 2021. Demon- strating the Evolution of GANs through t-SNE. In Applications of Evolutionary Computation: 24th International Conference, EvoApplications 2021, Held as Part of EvoStar 2021, Virtual Event, April 7–9,...

  8. [16]

    Victor Costa, Nuno Lourenço, and Penousal Machado. 2019. Coevolution of generative adversarial networks. In Applications of Evolutionary Computation: 22nd International Conference, EvoApplications 2019, Held as Part of EvoStar 2019, Leipzig, Germany, April 24–26, 2019, Proceed...

  9. [17]

    Li Deng. 2012. The mnist database of handwritten digit images for machine learning research [best of the web]. IEEE signal processing magazine 29, 6 (2012), 141–142

  10. [18]

    Diana Flores, Erik Hemberg, Jamal Toutouh, and Una-May O’Reily. 2022. Co- evolutionary generative adversarial networks for medical image augumentation at scale. In Proceedings of the genetic and evolutionary computation conference . 367–376

  11. [19]

    Ying Fu, Minxue Gong, Guang Yang, Hong Wei, and Jiliu Zhou. 2021. Evolutionary gan-based data augmentation for cardiac magnetic resonance image. Computers, Materials & Continua 68, 1 (2021), 1359–1374

  12. [20]

    Unai Garciarena, Roberto Santana, and Alexander Mendiburu. 2018. Evolved GANs for generating Pareto set approximations. In Proceedings of the genetic and evolutionary computation conference. 434–441

  13. [21]

    Ian Goodfellow, Jean Pouget-Abadie, Mehdi Mirza, Bing Xu, David Warde-Farley, Sherjil Ozair, Aaron Courville, and Yoshua Bengio. 2020. Generative adversarial networks. Commun. ACM 63, 11 (2020), 139–144

  14. [22]

    Juanjuan He, Qi Zhu, Kai Zhang, Piaoyao Yu, and Jinshan Tang. 2021. An evolvable adversarial network with gradient penalty for COVID-19 infection segmentation. Applied Soft Computing 113 (2021), 107947

  15. [23]

    Erik Hemberg, Jamal Toutouh, Abdullah Al-Dujaili, Tom Schmiedlechner, and Una-May O’Reilly. 2021. Spatial coevolution for generative adversarial network training. ACM Transactions on Evolutionary Learning and Optimization 1, 2 (2021), 1–28

  16. [24]

    Martin Heusel, Hubert Ramsauer, Thomas Unterthiner, Bernhard Nessler, and Sepp Hochreiter. 2017. GANs trained by a two time-scale update rule converge to a local nash equilibrium. Advances in neural information processing systems 30 (2017)

  17. [25]

    Philip Hingston and Mike Preuss. 2011. Red teaming with coevolution. In 2011 IEEE Congress of Evolutionary Computation (CEC) . IEEE, 1155–1163

  18. [26]

    David Keaveney and Colm O’Riordan. 2011. Evolving coordination for real-time strategy games. IEEE Transactions on Computational Intelligence and AI in Games 3, 2 (2011), 155–167

  19. [27]

    Krzysztof Krawiec and Malcolm Heywood. 2016. Solving Complex Problems with Coevolutionary Algorithms. InProceedings of the 2016 on Genetic and Evolutionary Computation Conference Companion. ACM, 687–713

  20. [28]

    Hyeeun Ku and Minhyeok Lee. 2023. TextControlGAN: Text-to-image synthesis with controllable generative adversarial networks. Applied Sciences 13, 8 (2023), 5098

  21. [29]

    Jerry Li, Aleksander Madry, John Peebles, and Ludwig Schmidt. 2018. On the limi- tations of first-order approximation in GAN dynamics. InInternational Conference on Machine Learning. PMLR, 3005–3013

  22. [30]

    Luke Metz, Ben Poole, David Pfau, and Jascha Sohl-Dickstein. 2016. Unrolled generative adversarial networks. arXiv preprint arXiv:1611.02163 (2016)

  23. [31]

    Jordan B Pollack and Alan D Blair. 1998. Co-evolution in the successful learning of backgammon strategy. Machine learning 32 (1998), 225–240

  24. [32]

    Paul Wiegand, and Edwin D

    Elena Popovici, Anthony Bucci, R. Paul Wiegand, and Edwin D. De Jong. 2012. Coevolutionary Principles. Springer Berlin Heidelberg, Berlin, Heidelberg, 987– 1033

  25. [33]

    Christopher D Rosin and Richard K Belew. 1997. New methods for competitive coevolution. Evolutionary Computation 5, 1 (1997), 1–29

  26. [34]

    George Rush, Daniel R Tauritz, and Alexander D Kent. 2015. Coevolutionary agent-based network defense lightweight event system (CANDLES). In Proceed- ings of the Companion Publication of the 2015 Annual Conference on Genetic and Evolutionary Computation. 859–866

  27. [35]

    Francisco Sedeño, Jamal Toutouh, and Francisco Chicano. 2025. Generate more than one child in your co-evolutionary semi-supervised learning GAN. In Pro- ceedings of the The Leading European Event on Bio -Inspired AI (EVOSTAR). 16. doi:10.5281/zenodo.15124788

  28. [36]

    Travis Service and Daniel Tauritz. 2009. Increasing infrastructure resilience through competitive coevolution. New Mathematics and Natural Computation 5, 02 (2009), 441–457

  29. [37]

    Karl Sims. 1994. Evolving 3D morphology and behavior by competition. Artificial life 1, 4 (1994), 353–372

  30. [38]

    Moshe Sipper. 2011. Evolved to Win. Lulu. com

  31. [39]

    Fanglei Sun, Qian Tao, Jianqiao Hu, and Jieqiong Liu. 2021. Composite evolu- tionary gan for natural language generation with temper control. In 2021 7th International Conference on Computer and Communications (ICCC) . IEEE, 1710– 1714

  32. [40]

    Jamal Toutouh, Mathias Esteban, and Sergio Nesmachnow. 2020. Paral- lel/distributed generative adversarial neural networks for data augmentation of COVID-19 training images. In Latin American High Performance Computing Conference. Springer, 162–177

  33. [41]

    Jamal Toutouh, Erik Hemberg, and Una-May O’Reilly. 2019. Spatial evolutionary generative adversarial networks. In Proceedings of the genetic and evolutionary computation conference. 472–480

  34. [42]

    Jamal Toutouh, Erik Hemberg, and Una-May O’Reilly. 2020. Analyzing the com- ponents of distributed coevolutionary gan training. In Parallel Problem Solving from Nature–PPSN XVI: 16th International Conference, PPSN 2020, Leiden, The Netherlands, September 5-9, 2020, Proceedings...

  35. [43]

    Jamal Toutouh, Erik Hemberg, and Una-May O’Reilly. 2020. Data dieting in gan training. In Deep Neural Evolution: Deep Learning with Evolutionary Computation . Springer, 379–400

  36. [44]

    Jamal Toutouh, Subhash Nalluru, Erik Hemberg, and Una-May O’Reilly. 2023. Semi-supervised generative adversarial networks with spatial coevolution for enhanced image generation and classification. Applied Soft Computing 148 (2023), 110890

  37. [45]

    Jamal Toutouh and Una-May O’Reilly. 2021. Signal propagation in a gradient- based and evolutionary learning system. In Proceedings of the Genetic and Evolu- tionary Computation Conference. 377–385

  38. [46]

    Vallender

    S.S. Vallender. 1974. Calculation of the Wasserstein distance between probability distributions on the line. Theory of Probability & Its Applications 18, 4 (1974), 784–786

  39. [47]

    Chaoyue Wang, Chang Xu, Xin Yao, and Dacheng Tao. 2019. Evolutionary generative adversarial networks. IEEE Transactions on Evolutionary Computation 23, 6 (2019), 921–934

  40. [48]

    Gabriel Wang, Anish Thite, Rodd Talebi, Anthony D’Achille, Alex Mussa, and Jason Zutty. 2022. Evolving SimGANs to improve abnormal electrocardiogram classification. In Proceedings of the Genetic and Evolutionary Computation Confer- ence Companion. 1887–1894

  41. [49]

    Josh L Wilkerson and Daniel Tauritz. 2010. Coevolutionary automated software correction. In Proceedings of the 12th annual conference on Genetic and evolutionary computation. 1391–1392

  42. [50]

    Guohao Ying, Xin He, Bin Gao, Bo Han, and Xiaowen Chu. 2022. EAGAN: Efficient two-stage evolutionary architecture search for GANs. In European conference on computer vision. Springer, 37–53

  43. [51]

    Hongke Zhao, Xinpeng Wu, Chuang Zhao, Lei Zhang, Haiping Ma, and Fan Cheng. 2021. Coea: A cooperative–competitive evolutionary algorithm for bidi- rectional recommendations. IEEE Transactions on Evolutionary Computation 26, 1 (2021), 28–42

  44. [52]

    Wenbo Zheng, Chao Gou, Lan Yan, and Fei-Yue Wang. 2019. Differential- evolution-based generative adversarial networks for edge detection. In Proceed- ings of the IEEE/CVF international conference on computer vision workshops . 0–0

Pith tools

Reviewed August 6, 2026 · model on record in the stance chip above.