Pith. sign in

REVIEW 4 major objections 6 minor 56 references

SiamNAS: Siamese Surrogate Model for Dominance Relation Prediction in Multi-objective Neural Architecture Search

T0 review · 4 major / 6 minor · reviewed 2026-08-07 · deepseek-v4-flash

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗

Pith's one-line read This paper claims that a Siamese-network surrogate that predicts pairwise dominance between architectures can replace true objective evaluation throughout an evolutionary multi-objective NAS search, finding the best NAS-Bench-201…

desk verdict A clean comparator idea for multi-objective NAS, but the efficiency claim is undermined by missing budget-matched baselines and a search space small enough to be covered by the evolutionary budget. read the letter →

arxiv 2506.02623 v1 pith:IMWZHIVY submitted 2025-06-03 cs.LG cs.AIcs.CV

Yuyang Zhou , Ferrante Neri , Yew-Soon Ong , Ruibin Bai This is my paper
classification cs.LGcs.AIcs.CV
keywords NeuralArchitectureSearchMulti-objectiveOptimizationSurrogateModelSiameseNetworkDominanceRelationParetoFrontNAS-Bench-201EvolutionaryAlgorithm
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 proposes SiamNAS, a multi-objective neural architecture search method built around a surrogate that compares two architectures and predicts which dominates the other, avoiding true evaluations during the search loop. The claim is that a lightweight ensemble of Siamese MLP blocks, trained on 600 pre-evaluated architectures from NAS-Bench-201, is accurate enough (about 92 percent) to drive tournament selection and non-dominated sorting. The framework replaces crowding distance with a simple bias toward larger models, then evaluates only the final front with true metrics. If correct, this makes multi-objective NAS dramatically cheaper: the paper reports finding the best CIFAR-10 architecture and the second-best ImageNet-16-120 architecture among all compared methods in about 0.01 GPU days.

What carries the argument

The load-bearing object is the Siamese surrogate ensemble. Each block shares weights between two MLP encoders, forms the difference vector d = b1 - b2, passes it through a two-layer MLP with sigmoid output, and rounds the result to a 0/1 dominance vote. An ensemble of seven such blocks makes a prediction, and a two-direction majority vote yields one of three outputs. This comparator replaces every dominance test in the evolutionary loop, including tournament selection and the efficient non-dominated sort, while the crowding-distance step is replaced by a heuristic that keeps larger models within a front.

What would settle it

Query the trained surrogate on a sample of triples (A, B, C) from NAS-Bench-201 and test whether A-vs-B, B-vs-C, and C-vs-A predictions form a cycle; a nontrivial cycle rate would show that Algorithm 3's front assignment is arbitrary, while a near-zero cycle rate would confirm the consistency assumption on which the method rests.

Watch

Extended reading notes

Core claim

The central discovery is that a pairwise dominance classifier can stand in for the expensive objective functions in evolutionary multi-objective NAS. Instead of predicting accuracy, parameter count, or FLOPs, the surrogate takes two one-hot encoded architectures, computes an embedding difference through shared MLPs, and outputs a rounded scalar indicating whether the first architecture dominates the second. An ensemble of such blocks and an asymmetric two-direction check produce three outcomes—first dominates, second dominates, or non-dominated—which feed directly into the non-dominated sorting of a modified NSGA-II-style loop. On NAS-Bench-201, the resulting SiamNAS found the theoretical best architecture for CIFAR-10 (5.63 percent test error) and near-best for CIFAR-100 and ImageNet-16-120, with all true evaluations confined to the initial 600 training architectures and the final front.

Load-bearing premise

The load-bearing premise is that the surrogate's pairwise dominance predictions are consistent enough—asymmetric and transitive, with no cycles—that non-dominated sorting over those predictions yields a meaningful front ordering.

Editorial extensions

If this is right

  • During the search phase no true objective values are needed; only 600 architectures are evaluated up front and a final front once, bounding total true evaluations by 650 with a population size of 50.
  • A surrogate trained on CIFAR-10 transfers to CIFAR-100 and ImageNet-16-120, yielding the second-best ImageNet error among all compared methods.
  • The runtime excluding true evaluations is about 0.01 GPU days, far below gradient-based methods (4 GPU days) and Bayesian-optimization-based neural predictors (2 GPU days).
  • Increasing the ensemble size beyond seven Siamese blocks gives diminishing returns, with prediction accuracy saturating around 92–93 percent.

Reading between the lines

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

  • The consistency assumption is left unverified: testing the trained surrogate for cycles on triples of architectures would directly show whether its pairwise dominance relation is transitive enough for non-dominated sorting to be meaningful.
  • The heuristic that favors larger models inside a front may itself drive much of the search quality; a controlled comparison against random survivor selection within a front would isolate the surrogate's contribution.
  • The same pairwise comparator could be applied to any multi-objective evolutionary loop with discrete encodings, not just NAS, whenever cheap pairwise dominance estimates are needed; the paper gestures toward this generality in its SOS discussion.
  • The proof-of-concept results are limited to NAS-Bench-201's 15,625 architectures; scaling to larger search spaces would require retesting the one-hot encoding and the 600-sample training budget.
Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

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

Referee Report

4 major / 6 minor

Summary. The paper proposes SiamNAS, a multi-objective neural architecture search framework in which an ensemble of Siamese MLP blocks is trained to predict pairwise dominance between candidate architectures. The surrogate is trained on 600 architectures from NAS-Bench-201 using true objective values, and is then used inside an NSGA-II-style evolutionary loop to perform tournament selection and non-dominated sorting, while survivor selection replaces crowding distance with a heuristic preference for architectures with more parameters. Experiments on NAS-Bench-201 report that SiamNAS finds the oracle CIFAR-10 architecture, obtains competitive results on CIFAR-100 and ImageNet-16-120, and runs in 0.01 GPU days when true evaluations are excluded.

Significance. If the central claim is established, a dominance-relation surrogate that needs only a few hundred true evaluations would be a useful ingredient for multi-objective NAS, and the idea of replacing objective-function approximation with pairwise comparison is interesting and clearly presented. The paper includes pseudocode for all components, ablation studies over ensemble size and training-set size, and comparison against 26 algorithms. However, the current evidence does not isolate the surrogate's contribution: the evolutionary budget is large enough to cover the entire search space many times, no budget-matched control is provided, and the surrogate's low F1 score raises doubts about the reliability of the non-dominated sorting that drives selection. These issues are fixable with additional experiments, so the work is a promising proof of concept rather than a fully supported claim as it stands.

major comments (4)
  1. [§5.2, Table 4, Algorithm 1] The central claim that the Siamese surrogate enables efficient search is not isolated by any budget-matched baseline. With N=50 and T=2000, Algorithm 1 generates roughly 100,000 offspring in a search space of 5^6=15,625 architectures, so the evolutionary loop can visit every valid architecture many times. The fact that the final population contains the oracle CIFAR-10 architecture could therefore be due to coverage of the entire space rather than to the learned dominance relation. I request control runs with the same 600 true evaluations and the same 100k cheap comparisons but (a) a random comparator, (b) a parameter-count-only comparator, and (c) an oracle comparator using true objectives; these controls determine whether the surrogate is doing the work.
  2. [Table 1, Algorithm 3 lines 1–2] The surrogate's positive-class F1 score is only 0.70–0.79 on all datasets (Table 1). Algorithm 3 performs non-dominated sorting by thresholding these pairwise predictions; because each pair is classified independently, the resulting relation need not be asymmetric or transitive. Cycles or intransitivities would make the front assignment in lines 1–2 arbitrary and could alter the final population. Please report the fraction of cyclic triples on a held-out set, and/or compare Algorithm 3's front assignment against true non-dominated sorting on the validation set. Without this, the search's selection pressure is not characterized.
  3. [Section 4.3, Algorithm 3 lines 8–12] The final survivor selection within the last non-dominated front is not made by the surrogate at all: it is a heuristic that keeps the architectures with the largest number of trainable parameters. This heuristic is justified in Section 4.3 only by the empirical observation that larger models often perform better. Because this tie-break is applied after every generation, the reported results are confounded between the surrogate and the parameter-count heuristic. An ablation that replaces the surrogate-based front assignment with a random ordering while keeping the parameter-count tie-break is needed to separate these two contributions.
  4. [§4.1, Algorithm 1 line 13, Table 4] The runtime comparison is incomplete: Table 4 reports 0.01 GPU days for SiamNAS, but that excludes Phase 1 and Phase 3 true-objective evaluations (up to N_s + N = 650 evaluations by the paper's own accounting in Section 4.1). On NAS-Bench-201 those evaluations are free lookups, but in a real NAS setting they are the dominant cost. The comparison with methods such as BANANAS and ReNAS therefore does not support the abstract's 'within 0.01 GPU days' claim, and the total true-evaluation budget should be reported consistently (600 in Table 3 vs. up to 650 in Algorithm 1).
minor comments (6)
  1. [Section 3.3] The notation x1⪯x2 is nonstandard and potentially confusing: in most multi-objective optimization texts x⪯y means y dominates x, whereas the paper writes 'x1⪯x2, i.e. x1 dominates x2'. Please define the symbol explicitly and use a consistent convention.
  2. [Section 5, Figure 4] Equation (2) writes f_e = 1−acc, which suggests a value in [0,1], but Figure 4 reports values such as 5.627 and 3.078×10^1, which are evidently percentages. Specify the units consistently throughout the text and figures.
  3. [Table 3] NASWOT [31] appears twice with different values (7.04±0.81 on CIFAR-10 and 7.19±0.99 on CIFAR-10); one of these rows is likely a different variant or a typo and should be corrected or labeled.
  4. [Algorithm 2, line 20] The line 'M ← build ensemble by trained Siamese surrogate models m1...m_R' refers to m_R, but the loop variable is r; this should be m_{N_m} or m_r for consistency.
  5. [Section 5.2] The 'Random Search [13]' baseline is cited to a surrogate-assisted NAS paper rather than to the original random search approach; please ensure the citation matches the method being compared.
  6. [Section 5.2, Table 3] The 'SiamNAS-transfer' experiment is not described in the experimental setup. Specify whether the same 600 CIFAR-10 architectures are used to train the surrogate, whether the search is then run on the target dataset, and whether the final Phase 3 evaluation uses the target dataset's true objectives.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: the reported SiamNAS result is an externally evaluated search outcome, not an artifact of the surrogate's construction.

full rationale

The paper's derivation chain is self-contained. The surrogate is trained on dominance labels obtained from true objective values of Ns=600 randomly sampled NAS-Bench-201 architectures (Algorithm 1 Phase 1, Algorithm 2), then used to compare candidates during the evolutionary search. The final non-dominated set is not taken from the surrogate's predictions; Phase 3 explicitly re-evaluates the final population with true objective function values before reporting the best architecture. Therefore the headline claim that SiamNAS found the oracle CIFAR-10 architecture is an empirical outcome of a true evaluation, not a label that was baked into the training set or an identity forced by construction. Hyperparameters Ns and Nm are chosen via ablation on the same benchmark, and the reported 92% surrogate accuracy is not described with a held-out split; these are experimental-validity and possible overfitting concerns, not circularity. The citations to prior work by overlapping authors ([16], [29]) are used as inspiration for the Siamese/ranking idea, not as a load-bearing uniqueness theorem or as a substitute for the present experiments. The skeptic's coverage argument (2000 generations x 50 individuals can visit most of the 15,625-architecture space) is a claim about missing controls, not about the derivation reducing to its inputs. No equation, fitted parameter, or self-citation chain exhibits the specific reduction required for a circularity finding.

Assumptions & free parameters 5 free parameters · 5 assumptions · 0 invented entities

The central claims rest on several domain assumptions about NAS-Bench-201 and the surrogate's reliability, plus a set of hyperparameters chosen by ablation. No new physical entities are introduced.

free parameters (5)
  • Ns (surrogate training sample size) = 600
    Chosen via ablation in Table 2; controls the number of true evaluations and surrogate accuracy.
  • Nm (ensemble size) = 7
    Chosen via ablation in Table 1; yields at least second-best accuracy and F1 score.
  • Population size and generations = N=50, T=2000
    Evolutionary budget; 2000 by 50 offspring covers the 15,625-architecture search space many times over, likely contributing to finding the global optimum.
  • Crossover and mutation rates = Rc=0.7, Rm=0.1
    Hand-set hyperparameters for the evolutionary operators; no sensitivity analysis provided.
  • Surrogate training hyperparameters = lr=0.001, batch=100, epochs=20
    Hand-set for Adam training of the Siamese blocks; not tuned per dataset.
assumptions (5)
  • domain assumption Within a non-dominated front, architectures with more parameters tend to have higher accuracy.
    Invoked in Algorithm 3 (lines 8-12) to replace crowding distance; not validated on NAS-Bench-201.
  • domain assumption Train accuracy is a reliable proxy for the test error objective f_e.
    The surrogate is trained on 600 architectures using train accuracy (Section 5) while the reported results are test error rates (Table 3).
  • domain assumption The learned pairwise dominance relation is consistent enough for non-dominated sorting.
    Non-dominated sorting (Algorithm 3) assumes an asymmetric, transitive dominance relation; the surrogate is a learned classifier with no consistency check.
  • domain assumption A 600-architecture random sample from NAS-Bench-201 is sufficient to train a generalizing dominance surrogate.
    Section 5.1 ablation shows accuracy plateaus near Ns=600; the search space has 15,625 architectures.
  • domain assumption The three objectives (error, params, FLOPs) and the one-hot encoding capture all relevant architecture trade-offs.
    Section 3.1 encoding and Eqs. (2)-(3); standard for NAS-Bench-201.

how reviews work

0 comments
Cite this review

Pith. "Pith review of SiamNAS: Siamese Surrogate Model for Dominance Relation Prediction in Multi-objective Neural Architecture Search." pith.science (2026). https://pith.science/paper/IMWZHIVY

@misc{pith2026250602623,
  author       = {Pith},
  title        = {Pith review of: SiamNAS: Siamese Surrogate Model for Dominance Relation Prediction in Multi-objective Neural Architecture Search},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/IMWZHIVY}},
  note         = {Machine review of arXiv:2506.02623}
}
read the original abstract

Modern neural architecture search (NAS) is inherently multi-objective, balancing trade-offs such as accuracy, parameter count, and computational cost. This complexity makes NAS computationally expensive and nearly impossible to solve without efficient approximations. To address this, we propose a novel surrogate modelling approach that leverages an ensemble of Siamese network blocks to predict dominance relationships between candidate architectures. Lightweight and easy to train, the surrogate achieves 92% accuracy and replaces the crowding distance calculation in the survivor selection strategy with a heuristic rule based on model size. Integrated into a framework termed SiamNAS, this design eliminates costly evaluations during the search process. Experiments on NAS-Bench-201 demonstrate the framework's ability to identify Pareto-optimal solutions with significantly reduced computational costs. The proposed SiamNAS identified a final non-dominated set containing the best architecture in NAS-Bench-201 for CIFAR-10 and the second-best for ImageNet, in terms of test error rate, within 0.01 GPU days. This proof-of-concept study highlights the potential of the proposed Siamese network surrogate model to generalise to multi-tasking optimisation, enabling simultaneous optimisation across tasks. Additionally, it offers opportunities to extend the approach for generating Sets of Pareto Sets (SOS), providing diverse Pareto-optimal solutions for heterogeneous task settings.

Figures

Figures reproduced from arXiv: 2506.02623 by the authors.

Figure 1
Figure 1. Encoding a network from the NAS-Bench-201 [PITH_FULL_IMAGE:figures/full_fig_p003_1.png] view at source ↗
Figure 2
Figure 2. Implementation of the Siamese network surrogate [PITH_FULL_IMAGE:figures/full_fig_p004_2.png] view at source ↗
Figure 3
Figure 3. Framework of SiamNAS: Green blocks represent the evolutionary operators, where offspring generation and biased [PITH_FULL_IMAGE:figures/full_fig_p006_3.png] view at source ↗
Figures from the paper (1 more)
Figure 4
Figure 4. Figure 4: The solutions in a Pareto front found by SiamNAS [PITH_FULL_IMAGE:figures/full_fig_p008_4.png]

Discussion (0). Sign in to comment.

Reference graph

Works this paper leans on

56 extracted references · 37 canonical work pages

  1. [1]

    Reinhard Booysen and Anna Sergevna Bosman. 2024. Multi-objective Evolution- ary Neural Architecture Search for Recurrent Neural Networks.Neural Processing Letters 56 (2024), 200. https://doi.org/10.1007/s11063-024-11659-0

    work page doi:10.1007/s11063-024-11659-0 2024

  2. [2]

    Sinee Chaiyarin, Napassorn Rojbundit, Panichanok Piyabenjarad, Pimpattra Limpitigranon, Siraprapa Wisitthipakdeekul, Pawaree Nonthasaen, and Paniti Achararit. 2024. Neural architecture search for medicine: A survey.Informatics in Medicine Unlocked 50 (2024), 101565. https://doi.org/10.1016/j.imu.2024.101565 SiamNAS: Siamese Surrogate Model for Dominance R...

    arXiv 2024

  3. [3]

    Han Xiang Choong, Yew-Soon Ong, Abhishek Gupta, Caishun Chen, and Ray Lim. 2023. Jack and Masters of all Trades: One-Pass Learning Sets of Model Sets From Large Pre-Trained Models. IEEE Computational Intelligence Magazine 18, 3 (2023), 29–40. https://doi.org/10.1109/MCI.2023.3277769

    work page arXiv 2023

  4. [4]

    Sumit Chopra, Raia Hadsell, and Yann LeCun. 2005. Learning a similarity metric discriminatively, with application to face verification. In Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition (CVPR) . 539–546

    work page 2005

  5. [5]

    Xiangxiang Chu, Xiaoxing Wang, Bo Zhang, Shun Lu, Xiaolin Wei, and Junchi Yan. 2021. DARTS-: Robustly stepping out of performance collapse without indicators. In International Conference on Learning Representations

    work page 2021

  6. [6]

    Xiangxiang Chu, Bo Zhang, and Ruijun Xu. 2021. FairNAS: Rethinking eval- uation fairness of weight sharing neural architecture search. In Proceedings of the IEEE/CVF International Conference on Computer Vision . 12219–12228. https://doi.org/10.1109/ICCV48922.2021.01202

    arXiv 2021

  7. [7]

    Meyarivan

    Kalyanmoy Deb, Amrit Pratap, Sameer Agarwal, and T. Meyarivan. 2002. A fast and elitist multiobjective genetic algorithm: NSGA-II. IEEE Transactions on Evolutionary Computation 6, 2 (2002), 182–197. https://doi.org/10.1109/4235. 996017

    doi:10.1109/4235 2002

  8. [8]

    Junwei Dong, Boyu Hou, Liang Feng, Huajin Tang, Kay Chen Tan, and Yew-Soon Ong. 2023. A Cell-Based Fast Memetic Algorithm for Automated Convolutional Neural Architecture Design. IEEE Transactions on Neural Networks and Learning Systems 34, 11 (2023), 9040–9053. https://doi.org/10.1109/TNNLS.2022.3155230

    arXiv 2023

Show all 56 references
  1. [9]

    Xuanyi Dong and Yi Yang. 2019. One-shot neural architecture search via self- evaluated template network. In Proceedings of the IEEE/CVF International Confer- ence on Computer Vision . 3680–3689. https://doi.org/10.1109/ICCV.2019.00378

    2019

  2. [10]

    Xuanyi Dong and Yi Yang. 2019. Searching for a robust neural architecture in four GPU hours. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition. 1761–1770. https://doi.org/10.1109/CVPR.2019.00186

    2019

  3. [11]

    Xuanyi Dong and Yi Yang. 2020. NAS-Bench-201: Extending the Scope of Re- producible Neural Architecture Search. In International Conference on Learning Representations (ICLR). https://openreview.net/forum?id=HJxyZkBKDr

    2020

  4. [12]

    Stefan Falkner, Aaron Klein, and Frank Hutter. 2018. BOHB: Robust and efficient hyperparameter optimization at scale. In Proceedings of the 35th International Conference on Machine Learning , Vol. 80. 1437–1446

    2018

  5. [13]

    Liang Fan and Handing Wang. 2023. Surrogate-assisted evolutionary neural architecture search with network embedding. Complex & Intelligent Systems 9, 3 (2023), 3313–3331. https://doi.org/10.1007/s40747-022-00929-w

    2023 doi

  6. [14]

    Bryson Greenwood and Tyler McDonnell. 2022. Surrogate-assisted neuroevo- lution. In Proceedings of the Genetic and Evolutionary Computation Conference (Boston, Massachusetts) (GECCO ’22). Association for Computing Machinery, New York, NY, USA, 1048–1056. https://doi.org/10.114...

    2022

  7. [15]

    Bicheng Guo, Tao Chen, Shibo He, Haoyu Liu, Lilin Xu, Peng Ye, and Jiming Chen. 2022. Generalized global ranking-aware neural architecture ranker for efficient image classifier search. In Proceedings of the 30th ACM International Conference on Multimedia. 3730–3741. https://do...

    2022

  8. [16]

    Pengcheng Jiang, Yu Xue, Ferrante Neri, and Mohamed Wahib. 2024. Surrogate- Assisted Evolutionary Neural Architecture Search with Isomorphic Training and Prediction. In Advanced Intelligent Computing Technology and Applications , De- Shuang Huang, Xiankun Zhang, and Yijie Pan ...

    2024

  9. [17]

    Kingma and Jimmy Ba

    Diederik P. Kingma and Jimmy Ba. 2015. Adam: A Method for Stochastic Opti- mization. In ICLR, Yoshua Bengio and Yann LeCun (Eds.)

    2015

  10. [18]

    Nikita Klyuchnikov, Ilya Trofimov, Ekaterina Artemova, Mikhail Salnikov, Maxim Fedorov, Alexander Filippov, and Evgeny Burnaev. 2022. NAS-Bench-NLP: Neural Architecture Search Benchmark for Natural Language Processing. IEEE Access 10 (2022), 45736–45747. https://doi.org/10.110...

    2022

  11. [19]

    Alex Krizhevsky. 2009. Learning Multiple Layers of Features from Tiny Images. https://api.semanticscholar.org/CorpusID:18268744

    2009

  12. [20]

    George Kyriakides and Konstantinos Margaritis. 2022. Evolving graph convolu- tional networks for neural architecture search.Neural Computing and Applications 34 (2022), 899–909. https://doi.org/10.1007/s00521-021-05979-8

    2022 doi

  13. [21]

    Liam Li and Ameet Talwalkar. 2020. Random search and reproducibility for neural architecture search. In Proceedings of the 35th Uncertainty in Artificial Intelligence Conference. 367–377

    2020

  14. [22]

    Yuhong Li, Cong Hao, Pan Li, Jinjun Xiong, and Deming Chen. 2021. Generic neural architecture search via regression. In Advances in Neural Information Processing Systems, Vol. 34. 20476–20490

    2021

  15. [23]

    Hanxiao Liu, Karen Simonyan, and Yiming Yang. 2019. DARTS: Differentiable Architecture Search. International Conference on Learning Representations (ICLR) (2019)

    2019

  16. [24]

    Jiao Liu, Yew Soon Ong, and Melvin Wong. 2024. Finding Sets of Pareto Sets in Real-World Scenarios – A Multitask Multiobjective Perspective. In2024 IEEE Con- gress on Evolutionary Computation (CEC). 1–8. https://doi.org/10.1109/CEC60901. 2024.10611967

    2024

  17. [25]

    Shiqing Liu, Haoyu Zhang, and Yaochu Jin. 2022. A survey on computationally efficient neural architecture search. Journal of Automation and Intelligence 1, 1 (2022), 100002. https://doi.org/10.1016/j.jai.2022.100002

    2022

  18. [26]

    Zhichao Lu, Ran Cheng, Shihua Huang, Haoming Zhang, Changxiao Qiu, and Fan Yang. 2023. Surrogate-Assisted Multiobjective Neural Architecture Search for Real-Time Semantic Segmentation. IEEE Transactions on Artificial Intelligence 4, 6 (2023), 1602–1615. https://doi.org/10.1109...

    2023

  19. [27]

    Zhichao Lu, Ran Cheng, Yaochu Jin, Kay Chen Tan, and Kalyanmoy Deb. 2024. Neural Architecture Search as Multiobjective Optimization Benchmarks: Problem Formulation and Performance Assessment. IEEE Transactions on Evolutionary Computation 28, 2 (2024), 323–337. https://doi.org/...

    2024

  20. [28]

    Zhichao Lu, Kalyanmoy Deb, Erik Goodman, Wolfgang Banzhaf, and Vishnu Naresh Boddeti. 2020. NSGANetV2: Evolutionary Multi-objective Surrogate-Assisted Neural Architecture Search. In Computer Vision – ECCV 2020 , Andrea Vedaldi, Horst Bischof, Thomas Brox, and Jan-Michael Frahm...

    2020

  21. [29]

    Kuangda Lyu, Hao Li, Maoguo Gong, Lining Xing, and A. K. Qin. 2024. Surrogate- Assisted Evolutionary Multiobjective Neural Architecture Search Based on Trans- fer Stacking and Knowledge Distillation. IEEE Transactions on Evolutionary Computation 28, 3 (2024), 608–622. https://...

    2024

  22. [30]

    Lianbo Ma, Nan Li, Guo Yu, Xiaoyu Geng, Shi Cheng, Xingwei Wang, Min Huang, and Yaochu Jin. 2024. Pareto-Wise Ranking Classifier for Multiobjective Evolutionary Neural Architecture Search. IEEE Transactions on Evolutionary Computation 28, 3 (2024), 570–581. https://doi.org/10....

    2024

  23. [31]

    Joseph Mellor, Jack Turner, Amos Storkey, and Elliot J Crowley. 2021. Neural architecture search without training. In Proceedings of the 38th International Conference on Machine Learning , Vol. 139. 7588–7598

    2021

  24. [32]

    Pereira, Iury B.A

    Gean T. Pereira, Iury B.A. Santos, Luís P.F. Garcia, Thierry Urruty, Muriel Visani, and André C.P.L.F. De Carvalho. 2023. Neural architecture search with inter- pretable meta-features and fast predictors.Information Sciences 649 (2023), 119642. https://doi.org/10.1016/j.ins.20...

    2023

  25. [33]

    Hieu Pham, Melody Guan, Barret Zoph, Quoc Le, and Jeff Dean. 2018. Efficient neural architecture search via parameters sharing. In Proceedings of the 35th International Conference on Machine Learning , Vol. 80. 4095–4104

    2018

  26. [34]

    E. Real, A. Aggarwal, Y. Huang, and Q. V. Le. 2017. Large-scale evolution of image classifiers. In Proceedings of the International Conference on Machine Learning (ICML). 2902–2911

    2017

  27. [35]

    E. Real, A. Aggarwal, Y. Huang, and Q. V. Le. 2019. Regularized evolution for image classifier architecture search. In Proceedings of the AAAI Conference on Artificial Intelligence. 587

    2019

  28. [36]

    Esteban Real, Alok Aggarwal, Yanping Huang, and Quoc V Le. 2019. Regularized evolution for image classifier architecture search. In Proceedings of the AAAI Conference on Artificial Intelligence, Vol. 33. 4780–4789. https://doi.org/10.1609/ aaai.v33i01.33014780

    2019

  29. [37]

    Esteban Real, Alok Aggarwal, Yanping Huang, and Quoc V Le. 2020. Regularized Evolution for Image Classifier Architecture Search. In Proceedings of the AAAI Conference on Artificial Intelligence, Vol. 33. 4780–4789

    2020

  30. [38]

    Michael Ruchte, Arber Zela, Julien Siems, Josif Grabocka, and Frank Hutter

  31. [39]

    Berg, and Li Fei-Fei

    Olga Russakovsky, Jia Deng, Hao Su, Jonathan Krause, Sanjeev Satheesh, Sean Ma, Zhiheng Huang, Andrej Karpathy, Aditya Khosla, Michael Bernstein, Alexander C. Berg, and Li Fei-Fei. 2015. ImageNet Large Scale Visual Recognition Challenge. arXiv:1409.0575 [cs.CV] https://arxiv.o...

    2015 arXiv

  32. [40]

    Jaeho Seong, Chaehyun Lee, and Dong Seog Han. 2022. Neural Architecture Search for Real-Time Driver Behavior Recognition. In 2022 International Confer- ence on Artificial Intelligence in Information and Communication (ICAIIC). 104–108. https://doi.org/10.1109/ICAIIC54071.2022.9722706

    2022

  33. [41]

    Seyed Mahdi Shariatzadeh, Mahmood Fathy, Reza Berangi, and Mohammad Shahverdy. 2023. A Survey on Multi-Objective Neural Architecture Search.ArXiv abs/2307.09099 (2023). https://api.semanticscholar.org/CorpusID:259951144

    2023 arXiv

  34. [42]

    Yen, and Yuqiao Liu

    Yanan Sun, Xian Sun, Yuhan Fang, Gary G. Yen, and Yuqiao Liu. 2021. A Novel Training Protocol for Performance Predictors of Evolutionary Neural Architec- ture Search Algorithms. IEEE Transactions on Evolutionary Computation 25, 3 (2021), 524–536. https://doi.org/10.1109/TEVC.2...

    2021

  35. [43]

    Chen Wei, Yiping Tang, Chuang Niu Chuang Niu, Haihong Hu, Yue Wang, and Jimin Liang. 2021. Self-supervised representation learning for evolutionary neural architecture search. IEEE Computational Intelligence Magazine 16, 3 (2021), 33–49. https://doi.org/10.1109/MCI.2021.3084415

    2021

  36. [44]

    Wei Wen, Hanxiao Liu, Yiran Chen, Hai Li, Gabriel Bender, and Pieter-Jan Kinder- mans. 2020. Neural Predictor for Neural Architecture Search. InComputer Vision – ECCV 2020, Andrea Vedaldi, Horst Bischof, Thomas Brox, and Jan-Michael Frahm (Eds.). Springer International Publish...

    2020

  37. [45]

    Colin White, Willie Neiswanger, and Yash Savani. 2021. BANANAS: Bayesian optimization with neural architectures for neural architecture search. In Pro- ceedings of the AAAI Conference on Artificial Intelligence , Vol. 35. 10293–10301. https://doi.org/10.1609/aaai.v35i12.17233

    2021 doi

  38. [46]

    Yen, Bing Xue, and Mengjie Zhang

    Xiangning Xie, Yuqiao Liu, Yanan Sun, Gary G. Yen, Bing Xue, and Mengjie Zhang. 2022. BenchENAS: A Benchmarking Platform for Evolutionary Neural Architecture Search. IEEE Transactions on Evolutionary Computation 26, 6 (2022), 1473–1485. https://doi.org/10.1109/TEVC.2022.314752...

    2022

  39. [48]

    Yixing Xu, Yunhe Wang, Kai Han, Yehui Tang, Shangling Jui, Chunjing Xu, and Chang Xu. 2021. ReNAS: Relativistic evaluation of neural architecture search. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR). 4409–4418. https://doi.org/10....

    2021

  40. [49]

    Yu Xue, Xiaolong Han, Ferrante Neri, Jiafeng Qin, and Danilo Pelusi. 2025. A Gradient-Guided Evolutionary Neural Architecture Search. IEEE Transactions on Neural Networks and Learning Systems 36, 3 (2025), 4345–4357. https://doi.org/ 10.1109/TNNLS.2024.3371432

    2025

  41. [50]

    Yu Xue, Weinan Tong, Ferrante Neri, Peng Chen, Tao Luo, Liangli Zhen, and Xiao Wang. 2024. Evolutionary Architecture Search for Generative Adversarial Net- works Based On Weight Sharing. IEEE Transactions on Evolutionary Computation (2024), 1–1. https://doi.org/10.1109/TEVC.20...

    2024

  42. [51]

    Shangshang Yang, Ye Tian, Xiaoshu Xiang, Shichen Peng, and Xingyi Zhang

  43. [52]

    Xingyi Zhang, Ye Tian, Ran Cheng, and Yaochu Jin. 2015. An Efficient Approach to Nondominated Sorting for Evolutionary Multiobjective Optimization. IEEE Transactions on Evolutionary Computation 19, 2 (2015), 201–213. https://doi.org/ 10.1109/TEVC.2014.2308305

    2015

  44. [53]

    Barret Zoph and Quoc V Le. 2017. Neural Architecture Search with Reinforcement Learning. In Proceedings of the International Conference on Learning Representa- tions (ICLR)

    2017

  45. [54]

    Barret Zoph and Quoc V Le. 2017. Neural architecture search with reinforcement learning. In International Conference on Learning Representations

    2017

  46. [55]

    Barret Zoph, Vijay Vasudevan, Jonathon Shlens, and Quoc V. Le. 2018. Learning transferable architectures for scalable image recognition. In Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition (CVPR) . 1–14

    2018

  47. [2020]

    https://github.com/automl/NASLib

    NASLib: A Modular and Flexible Neural Architecture Search Library. https://github.com/automl/NASLib

  48. [2022]

    IEEE Transactions on Cognitive and Developmental Systems 14, 4 (2022), 1778–1792

    Accelerating Evolutionary Neural Architecture Search via Multifidelity Evaluation. IEEE Transactions on Cognitive and Developmental Systems 14, 4 (2022), 1778–1792. https://doi.org/10.1109/TCDS.2022.3179482

    2022

Pith tools

Reviewed August 7, 2026 · model on record in the stance chip above.