Pith. sign in

REVIEW 3 major objections 6 minor 74 references

Knowledge-aware Evolutionary Graph Neural Architecture Search

T0 review · 3 major / 6 minor · reviewed 2026-08-12 · deepseek-v4-flash

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

Pith's one-line read Trained on the NAS-Bench-Graph knowledge base, KEGNAS generates and evaluates transfer architectures in a few GPU seconds and warm-starts a multi-objective evolutionary search, outperforming strong GNAS baselines by 4.27% to 11.54%…

desk verdict Real-world experiments don't test the transfer mechanism as described, but the NAS-Bench-Graph evidence supports the core idea; the paper deserves revision, not rejection. read the letter →

arxiv 2411.17339 v1 pith:VQTUO6EB submitted 2024-11-26 cs.NE cs.AIcs.LG

Chao Wang , Jiaxuan Zhao , Lingling Li , Licheng Jiao , Fang Liu , Xu Liu , Shuyuan Yang This is my paper · ORCID
classification cs.NEcs.AIcs.LG
keywords graphneuralarchitecturesearchevolutionarytransferoptimizationmulti-objectivedeepmulti-outputGaussianprocessNAS-Bench-Graphwarm-startalgorithmnetworkssurrogate-assisted
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

KEGNAS sets out to end the zero-knowledge state of graph neural architecture search (GNAS): instead of starting every new graph dataset from scratch, it reuses a public architecture-and-performance database, NAS-Bench-Graph, to seed the search. The paper's central claim is that this prior knowledge can be transferred cheaply through two components, a knowledge model that maps a dataset to candidate architectures and a deep multi-output Gaussian process that predicts accuracy and parameter count for those candidates. Only the non-dominated candidates are fed into a multi-objective evolutionary algorithm as its initial population. On eight NAS-Bench-Graph datasets and five real-world graph datasets, the paper reports that this warm start produces architectures with 4.27% higher average accuracy than the strongest evolutionary baseline and 11.54% higher than the strongest differentiable baseline, while the transfer-architecture generation and evaluation stage itself takes only a few GPU seconds. The importance, if true, is that expensive per-dataset architecture search can be accelerated by standing on accumulated benchmark knowledge rather than beginning each task in ignorance.

What carries the argument

The machinery is a two-model warm-start pipeline. A knowledge model, built on a DAG graph decoder, learns the mapping from a dataset's task-feature distribution (computed by a variational graph autoencoder) to Pareto-optimal architectures in NAS-Bench-Graph, so that a new dataset can quickly generate candidate architectures. A deep multi-output Gaussian process, a multi-output GP whose kernel is parameterized by a neural network over architecture and dataset encodings, predicts #Acc and #Params for each candidate on the new dataset without training the candidates. Fast non-dominated sorting on these predicted metrics selects the transfer architectures that initialize a multi-objective evolutionary algorithm, using SBX crossover and PM mutation. This placement matters: the two learned models only shape the initial population, leaving the underlying MOEA's evaluation, selection, and reproduction untouched, so the framework can wrap any MOEA.

What would settle it

Train the DMOGP and knowledge model on the eight NAS-Bench-Graph datasets, then on a held-out graph dataset compute the rank correlation between DMOGP-predicted and true #Acc and #Params; if the transfer architectures selected by the predicted Pareto front are no better than randomly sampled architectures when used to warm-start the same MOEA, the claimed transfer gain is not real.

Watch

Extended reading notes

Core claim

The paper's discovery is that a database of evaluated GNN architectures can be turned into a reusable prior for new graph tasks. After training once on NAS-Bench-Graph, KEGNAS uses a variational graph autoencoder to represent each dataset by a task-feature distribution, a DAG-decoder knowledge model to generate candidate transfer architectures for a new dataset, and a deep multi-output Gaussian process with learned deep kernels to predict each candidate's #Acc and #Params without training it. Fast non-dominated sorting selects the predicted Pareto candidates, and those architectures initialize an NSGA-II population that then continues the search normally. The reported result is that this pipeline outperforms six established GNAS baselines on NAS-Bench-Graph, and on five real-world datasets it exceeds the best surrogate-assisted evolutionary baseline by 4.27% average accuracy and the leading differentiable baseline by 11.54%, while the warm-start stage costs a few GPU seconds. Ablations attribute the gain to both the knowledge model and the DMOGP: removing the surrogate's selection or the knowledge transfer entirely reduces the hypervolume of the final architecture set.

Load-bearing premise

The paper's load-bearing premise is that knowledge of good architectures on the eight NAS-Bench-Graph datasets transfers to a new dataset: the knowledge model must generate candidates whose predicted metrics are accurate enough that the non-dominated warm-start population beats a random initial population.

Editorial extensions

If this is right

  • GNAS no longer needs to be a from-scratch process: any future task can inherit knowledge stored once from NAS-Bench-Graph, so a new dataset's search starts from a population already enriched with plausible Pareto architectures.
  • The warm-start step is cheap by design, since generating and evaluating 500 candidate transfer architectures costs only a few GPU seconds, an order of magnitude below the search cost of the MOEA itself.
  • Because the knowledge model and DMOGP are decoupled from the evolutionary loop, KEGNAS can be combined with any MOEA, not just NSGA-II, without disturbing existing evaluation strategies.
  • The ablation results imply that both learned components matter: randomly chosen transfer architectures and randomly initialized populations both produce lower-hypervolume architecture sets on NAS-Bench-Graph.
  • On real-world datasets, the same warm-started framework stays competitive or better than state-of-the-art differentiable and RL-based search methods while using much less search time than most evolutionary baselines.

Reading between the lines

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

  • The paper reports the transfer-architecture generation and evaluation time separately from the search time, but the real-world experiments in Section 5.2.1 build an additional Gaussian surrogate from trained transfer architectures before running NSGA-II; a careful reader should therefore treat the headline speedup as warm-start plus surrogate-assisted search, and expect the warm-start-only gain to
  • The same recipe could be applied to other public NAS benchmarks whose search spaces are DAG-encodable: train the knowledge model and DMOGP on the benchmark's evaluated architectures, then warm-start search on a new task, though the paper's own limitation discussion notes that heterogeneous search spaces and non-node-classification tasks would require new search spaces and training data.
  • A sharper test of the transfer hypothesis than the paper reports would be to measure DMOGP rank correlation on genuinely out-of-distribution datasets; if the surrogate's predictions are poor there, the warm start degrades toward random initialization, which is precisely the failure mode the ablation does not fully exclude.
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 / 6 minor

Summary. The paper proposes KEGNAS, a knowledge-aware evolutionary graph neural architecture search method. It trains a knowledge model and a deep multi-output Gaussian process (DMOGP) on the NAS-Bench-Graph knowledge base. For a new dataset, the knowledge model generates candidate transfer architectures, the DMOGP predicts their #Acc and #Params, and non-dominated candidates warm-start an NSGA-II search. Experiments on NAS-Bench-Graph with leave-one-out source tasks and on five real-world datasets report accuracy improvements over evolutionary and differentiable baselines; ablations show the knowledge model and DMOGP each contribute.

Significance. The idea of reusing NAS benchmark knowledge to warm-start multi-objective GNAS is timely and practically relevant, and the paper ships its code. The NAS-Bench-Graph leave-one-out experiments (Section 5.1, Tables 2 and 7) are clean: the source tasks exclude the target, and the ablation against NSGA-II and random-transfer warm starting isolates the contributions of the knowledge model and DMOGP. However, the real-world experiments (Section 5.2) evaluate a modified pipeline that trains all transfer architectures on the target and then uses a Gaussian surrogate; this weakens the central claim that KEGNAS swiftly generates top-performing architectures via the knowledge base. The paper is a solid candidate after the real-world experimental gap is addressed.

major comments (3)
  1. [Section 5.2.1 and Table 6] The real-world experiments do not evaluate the proposed Algorithm 2 as described. The paper states that "we modify KEGNAS-NSGAII... we first train all transfer architectures on real-world graph datasets. Subsequently, Gaussian models are constructed using the transfer architectures and their real performance metrics." Thus all 500 candidate transfer architectures are actually trained on the target dataset (pre-search cost 1.0K–32.0K GPU seconds in Table 6), and the DMOGP predictions are not used to evaluate the candidates. The "few GPU seconds" claim therefore covers only the generation and DMOGP inference sub-step, while the accuracy numbers in Table 5 are produced by a pipeline whose dominant cost is training the transfer architectures. This mismatch means the real-world experiments do not support the abstract's claim that KEGNAS "swiftly" generates top-performing architectures through the knowledge base.
  2. [Section 5.2.1 and Table 5] The real-world setup lacks a control that isolates the effect of prior knowledge. Since the Gaussian surrogate is built from the real performance of all 500 transfer architectures, the reported gains over CTFGNAS and SANE could in principle be obtained by any surrogate-assisted NSGA-II fed with 500 randomly sampled trained architectures. A control experiment running the same pipeline (train 500 randomly sampled architectures, fit the same Gaussian surrogate, run NSGA-II) is necessary to attribute the 4.27% and 11.54% improvements to the knowledge model and DMOGP rather than to the surrogate-assisted search or the larger number of target trainings. The ablation in Section 5.3 is performed only on NAS-Bench-Graph, so it does not close this gap.
  3. [Section 5.2.1 and Table 6] The real-world evaluation budgets are not matched across baselines. KEGNAS-NSGAII trains 500 transfer architectures plus retrains the selected architecture for 400 epochs, while Random trains 100 architectures and F2GNN/SANE use supernet-style training. Because the accuracy comparison in Table 5 is not conditioned on equal compute or an equal number of target model trainings, the reported percentage gains are not an apples-to-apples comparison. The authors should either match the number of trained architectures across methods or report accuracy as a function of total GPU time for all methods.
minor comments (6)
  1. [Section 2.2] The text says the macro space of NAS-Bench-Graph contains eight DAGs, but it then lists nine labels (A, B, C, D, E, F, G, H, I). Please correct the count or the list.
  2. [Section 2.5, Eq. (8)] The symbol Σ' is used both for the noise matrix in the mean formula and for the predicted covariance on the left-hand side; please use distinct notation for these two quantities.
  3. [Table 6] The Cornell row for Genetic-GNN reports "54.K", which appears to be a typo for "54.0K" or "54K"; also the Flickr KEGNAS row uses a lowercase 'k' in "32.0K" while other entries use uppercase.
  4. [Section 5.2.1] The Gaussian surrogate model used in the modified real-world experiments is not specified (kernel choice, hyperparameter fitting, validation strategy). Please provide these details for reproducibility.
  5. [Section 5.2.2] The sentence "generating and evaluating candidate transfer architectures requires only a few GPU seconds in total, which is negligible compared with the search cost of NSGAII" is misleading in context, because the pre-search cost in Table 6 is dominated by training the transfer architectures, not by the few seconds of DMOGP inference.
  6. [Sections 4.2 and 5.2.1] The paper does not report any measure of DMOGP predictive accuracy (e.g., rank correlation or RMSE) on held-out or real-world datasets; such a measurement would directly support the claim that transfer architectures are selected by accurate predicted metrics.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: transfer models are source-only and final results are real evaluations.

full rationale

The derivation chain is self-contained against the external NAS-Bench-Graph benchmark and does not reduce to its own inputs. The knowledge model and DMOGP are trained only on source tasks (leave-one-out on NAS-Bench-Graph, Section 5.1.1; on real-world datasets, on the eight NAS-Bench-Graph datasets, Section 5.2.1), so the target dataset's performance labels are not used to fit either model. On NAS-Bench-Graph, the transfer architectures are selected via DMOGP predictions, but NSGA-II then evaluates real #Acc/#Params from the benchmark, and the reported results (Table 2) come from those real evaluations, not from the surrogate. On real-world datasets, the paper admittedly trains the transfer architectures on the target and fits a Gaussian surrogate to their real metrics (Section 5.2.1), which is a standard surrogate-assisted search step rather than a normative circularity; the final accuracies in Table 5 are obtained by retraining the selected architectures from scratch, so no reported number is a surrogate prediction. The few-GPU-seconds claim applies to the knowledge-model/DMOGP sub-step, and the discrepancy between that sub-step and the full real-world pipeline is a scope-of-claim issue, not an equation-level circularity. Self-citations in the related-work sections are contextual and not load-bearing; no uniqueness theorem or ansatz is imported from the authors' prior work. Hence no circular step can be quoted as a reduction by construction.

Assumptions & free parameters 7 free parameters · 6 assumptions · 0 invented entities

The method relies on standard ML building blocks (VGAE, D-VAE decoder, deep kernel GP, NSGA-II) and on a domain assumption that source-task knowledge transfers to new datasets through VGAE task features. The main hand-chosen hyperparameters (Ns, Nc, Np, task feature dimension) shape the training data and search budget. No fundamentally new entities are postulated; the knowledge model and DMOGP are trained models, not newly invented physical or conceptual objects.

free parameters (7)
  • Ns (number of non-dominated fronts per source task) = 10
    Chosen by hand in Section 5.1.1 to define how much of the Pareto front of each source task is used for training; determines training set size for knowledge model and DMOGP.
  • Nc (number of candidate transfer architectures) = 500
    Chosen in Section 5.1.1; the number of architectures generated by the knowledge model for each new dataset.
  • Np (population size) = 25 (NAS-Bench-Graph), 20 (real-world)
    Set in Sections 5.1.1 and 5.2.1; sensitivity analysis in Fig. 8 shows modest effect.
  • Task feature dimension = 32
    Dimension of VGAE latent space used as task feature s; set in Section 5.1.1.
  • Training epochs, learning rate, batch size = 400, 0.001, 256
    Standard settings for the knowledge model and DMOGP, following prior NAS works [57,14].
  • Retraining epochs for searched architectures = 400
    Used to report final accuracy; chosen in Section 5.2.1.
  • Gaussian surrogate model details = unspecified
    On real-world datasets, Gaussian models are built from 500 trained transfer architectures to replace evaluation in NSGA-II; implementation details are not given.
assumptions (6)
  • domain assumption The task family assumption: a new dataset's optimal architectures are related to source datasets through a shared task feature space S, so a decoder ptheta(alpha|s) trained on source tasks can generate competitive candidates for a new task.
    Introduced in Section 4.1, Eqs. (13)-(15). This is the core transfer premise; if false, warm-starting cannot help.
  • domain assumption VGAE-derived task features p(s|Db) capture dataset similarity relevant to architecture performance.
    Section 4.1; the entire transfer mechanism conditions on these embeddings, but no evidence links VGAE latent distance to accuracy similarity.
  • domain assumption The DMOGP surrogate, trained on source tasks with deep kernels, can predict #Acc and #Params on new datasets accurately enough for ranking.
    Section 4.2, Eqs. (18)-(19); prediction error on target datasets is never measured.
  • ad hoc to paper Top Ns=10 non-dominated fronts of source tasks contain sufficient and diverse knowledge.
    Algorithm 1, Line 5; no justification or sensitivity analysis for Ns.
  • standard math Standard ML and GP background: VGAE, GRU-based DAG decoder (D-VAE), deep kernel GP, NSGA-II, fast non-dominated sorting.
    Used throughout Sections 2 and 4; taken from prior literature [66,67,65,62].
  • ad hoc to paper In real-world experiments, Gaussian surrogates fitted on 500 transfer architectures are accurate enough to guide NSGA-II.
    Section 5.2.1; this replaces the true evaluation in the MOEA loop and is not validated.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Knowledge-aware Evolutionary Graph Neural Architecture Search." pith.science (2026). https://pith.science/paper/VQTUO6EB

@misc{pith2026241117339,
  author       = {Pith},
  title        = {Pith review of: Knowledge-aware Evolutionary Graph Neural Architecture Search},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/VQTUO6EB}},
  note         = {Machine review of arXiv:2411.17339}
}
read the original abstract

Graph neural architecture search (GNAS) can customize high-performance graph neural network architectures for specific graph tasks or datasets. However, existing GNAS methods begin searching for architectures from a zero-knowledge state, ignoring the prior knowledge that may improve the search efficiency. The available knowledge base (e.g. NAS-Bench-Graph) contains many rich architectures and their multiple performance metrics, such as the accuracy (#Acc) and number of parameters (#Params). This study proposes exploiting such prior knowledge to accelerate the multi-objective evolutionary search on a new graph dataset, named knowledge-aware evolutionary GNAS (KEGNAS). KEGNAS employs the knowledge base to train a knowledge model and a deep multi-output Gaussian process (DMOGP) in one go, which generates and evaluates transfer architectures in only a few GPU seconds. The knowledge model first establishes a dataset-to-architecture mapping, which can quickly generate candidate transfer architectures for a new dataset. Subsequently, the DMOGP with architecture and dataset encodings is designed to predict multiple performance metrics for candidate transfer architectures on the new dataset. According to the predicted metrics, non-dominated candidate transfer architectures are selected to warm-start the multi-objective evolutionary algorithm for optimizing the #Acc and #Params on a new dataset. Empirical studies on NAS-Bench-Graph and five real-world datasets show that KEGNAS swiftly generates top-performance architectures, achieving 4.27% higher accuracy than advanced evolutionary baselines and 11.54% higher accuracy than advanced differentiable baselines. In addition, ablation studies demonstrate that the use of prior knowledge significantly improves the search performance.

Figures

Figures reproduced from arXiv: 2411.17339 by the authors.

Figure 1
Figure 1. An illustrative example of the GCN for NAS-Bench-Graph. The architecture [PITH_FULL_IMAGE:figures/full_fig_p007_1.png] view at source ↗
Figure 2
Figure 2. Overview of KEGNAS, which consists of two phases: training and search. A [PITH_FULL_IMAGE:figures/full_fig_p012_2.png] view at source ↗
Figure 3
Figure 3. Overview of the knowledge model. where S = {s 1 , ..., s K} denotes a task space that contains K source tasks and s is the task feature describing the source task in T . In NAS-Bench-Graph, each NAS task corresponds to a unique benchmark dataset. The task feature s can be extracted from the benchmark dataset Db using graph embedding methods. Assuming that the task features s follow a probability distribution p(s|Db)… view at source ↗
Figures from the paper (5 more)
Figure 4
Figure 4. Figure 4: Flowchart of the DMOGP. 4.2. DMOGP According to the formula (11), the evaluation process of each architec￾ture must complete a lower-level optimization that involves multiple costly iterations of gradient descent over multiple epochs. Therefore, directly evalu￾ating a …
Figure 5
Figure 5. Figure 5: Similarity across different tasks using three metrics. [PITH_FULL_IMAGE:figures/full_fig_p020_5.png]
Figure 6
Figure 6. Figure 6: Comparison of the front (#Acc and #Params) obtained by KEGNAS-NSGAII [PITH_FULL_IMAGE:figures/full_fig_p024_6.png]
Figure 7
Figure 7. Figure 7: Frequency of the macro space and candidate operations in the real Pareto [PITH_FULL_IMAGE:figures/full_fig_p026_7.png]
Figure 8
Figure 8. Figure 8: Average HV values with various configurations ([10 [PITH_FULL_IMAGE:figures/full_fig_p033_8.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

74 extracted references · 37 canonical work pages

  1. [1]

    J. Chen, B. Li, K. He, Neighborhood convolutional graph neu- ral network, Knowledge-Based Systems 295 (2024) 111861. doi:https://doi.org/10.1016/j.knosys.2024.111861. URL https://www.sciencedirect.com/science/article/pii/ S0950705124004957

    arXiv 2024

  2. [2]

    T. N. Kipf, M. Welling, Semi-supervised classification with graph con- volutional networks, in: International Conference on Learning Repre- sentations, 2017

    2017

  3. [3]

    Hamilton, Z

    W. Hamilton, Z. Ying, J. Leskovec, Inductive representation learning on large graphs, in: Advances in Neural Information Processing Systems, Vol. 30, Curran Associates, Inc., 2017

    work page 2017

  4. [4]

    Veliˇ ckovi´ c, G

    P. Veliˇ ckovi´ c, G. Cucurull, A. Casanova, A. Romero, P. Li` o, Y. Bengio, Graph attention networks, in: International Conference on Learning Representations, 2018

    work page 2018

  5. [5]

    K. Xu, W. Hu, J. Leskovec, S. Jegelka, How powerful are graph neural networks?, in: International Conference on Learning Representations, 2019

    2019

  6. [6]

    Z. Wu, S. Pan, F. Chen, G. Long, C. Zhang, P. S. Yu, A comprehensive survey on graph neural networks, IEEE Transactions on Neural Net- works and Learning Systems 32 (1) (2021) 4–24. doi:10.1109/TNNLS. 2020.2978386

    arXiv 2021

  7. [7]

    X. Wang, W. Zhu, Automated machine learning on graph, in: Proceed- ings of the 27th ACM SIGKDD Conference on Knowledge Discovery & Data Mining, ACM, 2021, pp. 4082–4083. 35

    work page 2021

  8. [8]

    Y. Gao, P. Zhang, H. Yang, C. Zhou, Y. Hu, Z. Tian, Z. Li, J. Zhou, Graphnas++: Distributed architecture search for graph neural net- works, IEEE Transactions on Knowledge and Data Engineering 35 (7) (2023) 6973–6987. doi:10.1109/TKDE.2022.3178153

    arXiv 2023

Show all 74 references
  1. [9]

    J. Chen, J. Gao, Y. Chen, B. M. Oloulade, T. Lyu, Z. Li, Auto-gnas: A parallel graph neural architecture search framework, IEEE Transactions on Parallel and Distributed Systems 33 (11) (2022) 3117–3128. doi: 10.1109/TPDS.2022.3151895

    2022

  2. [10]

    L. Wei, H. Zhao, Z. He, Designing the topology of graph neural networks: A novel feature fusion perspective, in: Proceedings of the ACM Web Conference 2022, WWW ’22, ACM, 2022, p. 1381–1391. doi:10.1145/ 3485447.3512185

    2022

  3. [11]

    B. G.M., G. Pillai, Sequential node search for faster neural ar- chitecture search, Knowledge-Based Systems 300 (2024) 112145. doi:https://doi.org/10.1016/j.knosys.2024.112145. URL https://www.sciencedirect.com/science/article/pii/ S0950705124007792

    2024

  4. [12]

    Y. An, C. Zhang, X. Zheng, Knowledge reconstruction as- sisted evolutionary algorithm for neural network architec- ture search, Knowledge-Based Systems 264 (2023) 110341. doi:https://doi.org/10.1016/j.knosys.2023.110341. URL https://www.sciencedirect.com/science/article/pii/ S...

    2023

  5. [13]

    H. Zhao, Q. Yao, W. Tu, Search to aggregate neighborhood for graph neural network, in: IEEE International Conference on Data Engineering, IEEE, 2021, pp. 552–563. doi:10.1109/ICDE51399.2021.00054

    2021

  6. [14]

    Shala, T

    G. Shala, T. Elsken, F. Hutter, J. Grabocka, Transfer NAS with meta- learned bayesian surrogates, in: International Conference on Learning Representations, 2023

    2023

  7. [15]

    M. Shi, Y. Tang, X. Zhu, Y. Huang, D. Wilson, Y. Zhuang, J. Liu, Genetic-gnn: Evolutionary architecture search for graph neural net- works, Knowledge-Based Systems 247 (2022) 108752. doi:https: //doi.org/10.1016/j.knosys.2022.108752. 36

    2022

  8. [16]

    C. Wang, L. Jiao, J. Zhao, L. Li, X. Liu, F. Liu, S. Yang, Bi-level multiobjective evolutionary learning: A case study on multitask graph neural topology search, IEEE Transactions on Evolutionary Computa- tion 28 (1) (2024) 208–222. doi:10.1109/TEVC.2023.3255263

    2024

  9. [17]

    Y. Gao, P. Zhang, C. Zhou, H. Yang, Z. Li, Y. Hu, P. S. Yu, Hgnas++: Efficient architecture search for heterogeneous graph neural networks, IEEE Transactions on Knowledge and Data Engineering 35 (9) (2023) 9448–9461. doi:10.1109/TKDE.2023.3239842

    2023

  10. [18]

    K. C. Tan, L. Feng, M. Jiang, Evolutionary transfer optimization - a new frontier in evolutionary computation research, IEEE Computational Intelligence Magazine 16 (1) (2021) 22–33. doi:10.1109/MCI.2020. 3039066

    2021 doi

  11. [20]

    Shakeri, E

    M. Shakeri, E. Miahi, A. Gupta, Y.-S. Ong, Scalable transfer evolution- ary optimization: Coping with big task instances, IEEE Transactions on Cybernetics 53 (10) (2023) 6160–6172. doi:10.1109/TCYB.2022. 3164399

    2023 doi

  12. [21]

    Y. Qin, Z. Zhang, X. Wang, Z. Zhang, W. Zhu, NAS-bench-graph: Benchmarking graph neural architecture search, in: Thirty-sixth Con- ference on Neural Information Processing Systems Datasets and Bench- marks Track, 2022

    2022

  13. [22]

    Gilmer, S

    J. Gilmer, S. S. Schoenholz, P. F. Riley, O. Vinyals, G. E. Dahl, Neural message passing for quantum chemistry, in: Proceedings of the 34th International Conference on Machine Learning, Vol. 70, PMLR, 2017, pp. 1263–1272

    2017

  14. [23]

    Defferrard, X

    M. Defferrard, X. Bresson, P. Vandergheynst, Convolutional neural net- works on graphs with fast localized spectral filtering, in: Advances in Neural Information Processing Systems, Vol. 29, Curran Associates, Inc., 2016. 37

    2016

  15. [24]

    F. M. Bianchi, D. Grattarola, L. Livi, C. Alippi, Graph neural networks with convolutional arma filters, IEEE Transactions on Pattern Analysis and Machine Intelligence 44 (7) (2022) 3496–3507.doi:10.1109/TPAMI. 2021.3054830

    2022

  16. [25]

    Morris, M

    C. Morris, M. Ritzert, M. Fey, W. L. Hamilton, J. E. Lenssen, G. Rattan, M. Grohe, Weisfeiler and leman go neural: Higher-order graph neural networks, Proceedings of the AAAI Conference on Artificial Intelligence 33 (01) (2019) 4602–4609. doi:10.1609/aaai.v33i01.33014602

    2019 doi

  17. [26]

    Jiang, P

    S. Jiang, P. Balaprakash, Graph neural network architecture search for molecular property prediction, in: IEEE International Conference on Big Data, IEEE, 2020, pp. 1346–1353. doi:10.1109/BigData50022. 2020.9378060

    2020

  18. [27]

    G. Li, M. Muller, A. Thabet, B. Ghanem, DeepGCNs: Can GCNs Go As Deep As CNNs? , in: 2019 IEEE/CVF International Conference on Computer Vision (ICCV), IEEE Computer Society, Los Alamitos, CA, USA, 2019, pp. 9266–9275. doi:10.1109/ICCV.2019.00936. URL https://doi.ieeecomputerso...

    2019

  19. [28]

    H. Liu, K. Simonyan, Y. Yang, DARTS: Differentiable architecture search, in: International Conference on Learning Representations, 2019

    2019

  20. [29]

    Y. Zhao, D. Wang, X. Gao, R. Mullins, P. Lio, M. Jamnik, Prob- abilistic dual network architecture search on graphs, arXiv preprint arXiv:2003.09676 (2020)

    2020 arXiv

  21. [30]

    Y. Qin, X. Wang, Z. Zhang, H. Chen, W. Zhu, Multi-task graph neural architecture search with task-aware collaboration and curriculum, in: Thirty-seventh Conference on Neural Information Processing Systems, 2023. URL https://openreview.net/forum?id=TOxpAwp0VE

    2023

  22. [31]

    B. Zoph, Q. Le, Neural architecture search with reinforcement learning, in: International Conference on Learning Representations, 2017

    2017

  23. [32]

    K. Zhou, X. Huang, Q. Song, R. Chen, X. Hu, Auto-gnn: Neural archi- tecture search of graph neural networks, Frontiers in big Data 5 (2022) 1029307. 38

    2022

  24. [33]

    Y. Gao, H. Yang, P. Zhang, C. Zhou, Y. Hu, Graph neural architecture search, in: Proceedings of the Twenty-Ninth International Joint Confer- ence on Artificial Intelligence, IJCAI-20, International Joint Conferences on Artificial Intelligence Organization, 2020, pp. 1403–1409,...

    2020 doi

  25. [34]

    J. Cai, X. Wang, H. Li, Z. Zhang, W. Zhu, Multimodal graph neural architecture search under distribution shifts, Proceedings of the AAAI Conference on Artificial Intelligence 38 (8) (2024) 8227–8235. doi:10. 1609/aaai.v38i8.28663. URL https://ojs.aaai.org/index.php/AAAI/articl...

    2024

  26. [35]

    Yao, Evolving artificial neural networks, Proceedings of the IEEE 87 (9) (1999) 1423–1447

    X. Yao, Evolving artificial neural networks, Proceedings of the IEEE 87 (9) (1999) 1423–1447. doi:10.1109/5.784219

    1999 doi

  27. [36]

    Y. Liu, Y. Sun, B. Xue, M. Zhang, G. G. Yen, K. C. Tan, A survey on evolutionary neural architecture search, IEEE Transactions on Neural Networks and Learning Systems 34 (2) (2023) 550–570. doi:10.1109/ TNNLS.2021.3100554

    2023

  28. [37]

    Z. Lu, R. Cheng, Y. Jin, K. C. Tan, K. Deb, Neural architecture search as multiobjective optimization benchmarks: Problem formulation and performance assessment, IEEE Transactions on Evolutionary Computa- tion 28 (2) (2024) 323–337. doi:10.1109/TEVC.2022.3233364

    2024

  29. [38]

    Nunes, P

    M. Nunes, P. M. Fraga, G. L. Pappa, Fitness landscape analysis of graph neural network architecture search spaces, in: Proceedings of the Genetic and Evolutionary Computation Conference, GECCO ’21, ACM, 2021, p. 876–884. doi:10.1145/3449639.3459318

    2021

  30. [39]

    Y. Liu, J. Liu, A surrogate evolutionary neural architecture search al- gorithm for graph neural networks, Applied Soft Computing 144 (2023) 110485. doi:https://doi.org/10.1016/j.asoc.2023.110485. URL https://www.sciencedirect.com/science/article/pii/ S1568494623005033

    2023

  31. [40]

    C. Wang, J. Zhao, L. Li, L. Jiao, F. Liu, S. Yang, Automatic graph topology-aware transformer, IEEE Transactions on Neural Networks and Learning Systems (2024) 1–15 doi:10.1109/TNNLS.2024.3440269. 39

    2024

  32. [41]

    X. Xue, C. Yang, L. Feng, K. Zhang, L. Song, K. C. Tan, How to Exploit Optimization Experience? Revisiting Evolutionary Sequential Transfer Optimization: Part A - Benchmark Problems, techrxiv preprint techrxiv.21694754.v1 (2022). doi:10.36227/techrxiv.21694754.v1

    2022 doi

  33. [42]

    Gupta, Y.-S

    A. Gupta, Y.-S. Ong, L. Feng, Multifactorial evolution: Toward evolu- tionary multitasking, IEEE Transactions on Evolutionary Computation 20 (3) (2016) 343–357. doi:10.1109/TEVC.2015.2458037

    2016

  34. [43]

    X. Ma, M. Xu, Y. Yu, H. Liu, Y. Wang, L. Wang, Y. Qi, J. Xiong, Enhancing evolutionary multitasking optimization by leveraging inter-task knowledge transfers and improved evolu- tionary operators, Knowledge-Based Systems 259 (2023) 110027. doi:https://doi.org/10.1016/j.knosys....

    2023

  35. [44]

    F. Gao, W. Gao, L. Huang, S. Zhang, M. Gong, L. Wang, Effective transferred knowledge identified by bipartite graph for multiobjective multitasking optimization, Knowledge-Based Systems 290 (2024) 111530. doi:https://doi.org/10.1016/j.knosys.2024.111530. URL https://www.scienc...

    2024

  36. [45]

    C. Wang, J. Zhao, L. Li, L. Jiao, J. Liu, K. Wu, A multi-transformation evolutionary framework for influence maximization in social networks, IEEE Computational Intelligence Magazine 18 (1) (2023) 52–67. doi: 10.1109/MCI.2022.3222050

    2023

  37. [46]

    X. Xue, C. Yang, L. Feng, K. Zhang, L. Song, K. C. Tan, How to Exploit Optimization Experience? Revisiting Evolutionary Sequential Transfer Optimization: Part B - Algorithm Analysis, techrxiv preprint techrxiv.21694832.v1 (2022). doi:10.36227/techrxiv.21694832.v1

    2022 doi

  38. [47]

    C. Wang, J. Liu, K. Wu, Z. Wu, Solving multitask optimization prob- lems with adaptive knowledge transfer via anomaly detection, IEEE Transactions on Evolutionary Computation 26 (2) (2022) 304–318. doi:10.1109/TEVC.2021.3068157. 40

    2022

  39. [48]

    C. Wang, K. Wu, J. Liu, Evolutionary multitasking auc optimization [research frontier], IEEE Computational Intelligence Magazine 17 (2) (2022) 67–82. doi:10.1109/MCI.2022.3155325

    2022

  40. [49]

    X. Wang, Y. Jin, S. Schmitt, M. Olhofer, R. Allmendinger, Transfer learning based surrogate assisted evolutionary bi- objective optimization for objectives with different evalua- tion times, Knowledge-Based Systems 227 (2021) 107190. doi:https://doi.org/10.1016/j.knosys.2021.1...

    2021

  41. [50]

    X. Wang, Y. Jin, S. Schmitt, M. Olhofer, Transfer learning based co- surrogate assisted evolutionary bi-objective optimization for objectives with non-uniform evaluation times, Evolutionary Computation 30 (2) (2022) 221–251. doi:10.1162/evco_a_00300

    2022 doi

  42. [51]

    X. Xue, C. Yang, Y. Hu, K. Zhang, Y.-M. Cheung, L. Song, K. C. Tan, Evolutionary sequential transfer optimization for objective- heterogeneous problems, IEEE Transactions on Evolutionary Computa- tion 26 (6) (2022) 1424–1438. doi:10.1109/TEVC.2021.3133874

    2022

  43. [52]

    Nomura, S

    M. Nomura, S. Watanabe, Y. Akimoto, Y. Ozaki, M. Onishi, Warm starting cma-es for hyperparameter optimization, Proceedings of the AAAI Conference on Artificial Intelligence 35 (10) (2021) 9188–9196. doi:10.1609/aaai.v35i10.17109

    2021 doi

  44. [53]

    Zhang, A

    N. Zhang, A. Gupta, Z. Chen, Y.-S. Ong, Multitask neuroevolution for reinforcement learning with long and short episodes, IEEE Trans- actions on Cognitive and Developmental Systems (2022) 1–1 doi:10. 1109/TCDS.2022.3221805

    2022

  45. [54]

    Feng, Y.-S

    L. Feng, Y.-S. Ong, S. Jiang, A. Gupta, Autoencoding evolutionary search with learning across heterogeneous problems, IEEE Transactions on Evolutionary Computation 21 (5) (2017) 760–772. doi:10.1109/ TEVC.2017.2682274

    2017

  46. [55]

    X. Xue, C. Yang, L. Feng, K. Zhang, L. Song, K. C. Tan, Solution transfer in evolutionary optimization: An empirical study on sequential 41 transfer, IEEE Transactions on Evolutionary Computation (2023) 1– 1doi:10.1109/TEVC.2023.3339506

    2023

  47. [56]

    E. O. Scott, K. A. De Jong, First complexity results for evolutionary knowledge transfer, in: Proceedings of the 17th ACM/SIGEVO Con- ference on Foundations of Genetic Algorithms, FOGA ’23, Associa- tion for Computing Machinery, New York, NY, USA, 2023, p. 140–151. doi:10.1145...

    2023

  48. [57]

    H. Lee, E. Hyung, S. J. Hwang, Rapid neural architecture search by learning to generate graphs from datasets, in: International Conference on Learning Representations, 2021

    2021

  49. [58]

    Z. Lu, G. Sreekumar, E. Goodman, W. Banzhaf, K. Deb, V. N. Boddeti, Neural architecture transfer, IEEE Transactions on Pattern Analysis and Machine Intelligence 43 (9) (2021) 2971–2989.doi:10.1109/TPAMI. 2021.3052758

    2021

  50. [59]

    X. Wang, Z. Zhang, W. Zhu, Automated graph machine learning: Approaches, libraries and directions, arXiv preprint arXiv:2201.01288 (2022)

    2022 arXiv

  51. [60]

    Astudillo, P

    R. Astudillo, P. Frazier, Bayesian optimization of composite functions, in: Proceedings of the 36th International Conference on Machine Learn- ing, Vol. 97, PMLR, 2019, pp. 354–363

    2019

  52. [61]

    J. Luo, J. Feng, R. Jin, A new approach to building the gaussian process model for expensive multi-objective optimization, in: 2019 9th Interna- tional Conference on Information Science and Technology, IEEE, 2019, pp. 374–379. doi:10.1109/ICIST.2019.8836854

    2019

  53. [62]

    H. Liu, J. Cai, Y.-S. Ong, Remarks on multi-output gaussian process regression, Knowledge-Based Systems 144 (2018) 102–121

    2018

  54. [63]

    W. J. Maddox, M. Balandat, A. G. Wilson, E. Bakshy, Bayesian opti- mization with high-dimensional outputs, in: Advances in Neural Infor- mation Processing Systems, Vol. 34, Curran Associates, Inc., 2021

    2021

  55. [64]

    Jakkala, Deep gaussian processes: A survey, arXiv preprint arXiv:2106.12135 (2021)

    K. Jakkala, Deep gaussian processes: A survey, arXiv preprint arXiv:2106.12135 (2021). 42

    2021 arXiv

  56. [65]

    K. Deb, A. Pratap, S. Agarwal, T. Meyarivan, A fast and elitist multiob- jective genetic algorithm: Nsga-ii, IEEE Transactions on Evolutionary Computation 6 (2) (2002) 182–197. doi:10.1109/4235.996017

    2002

  57. [66]

    T. N. Kipf, M. Welling, Variational graph auto-encoders, arXiv preprint arXiv:1611.07308 (2016)

    2016 arXiv

  58. [67]

    Zhang, S

    M. Zhang, S. Jiang, Z. Cui, R. Garnett, Y. Chen, D-vae: A variational autoencoder for directed acyclic graphs, in: Advances in Neural Infor- mation Processing Systems, Vol. 32, Curran Associates, Inc., 2019

    2019

  59. [68]

    Y. Zhu, Y. Du, Y. Wang, Y. Xu, J. Zhang, Q. Liu, S. Wu, A survey on deep graph generation: Methods and applications, in: B. Rieck, R. Pas- canu (Eds.), Proceedings of the First Learning on Graphs Conference, Vol. 198 of Proceedings of Machine Learning Research, PMLR, 2022, pp....

    2022

  60. [69]

    L. Li, A. Talwalkar, Random search and reproducibility for neural ar- chitecture search, in: Proceedings of The 35th Uncertainty in Artificial Intelligence Conference, Vol. 115, PMLR, 2020, pp. 367–377

    2020

  61. [70]

    E. Real, A. Aggarwal, Y. Huang, Q. V. Le, Regularized evolution for image classifier architecture search, Proceedings of the AAAI Conference on Artificial Intelligence 33 (01) (2019) 4780–4789

    2019

  62. [71]

    K. Xu, C. Li, Y. Tian, T. Sonobe, K.-i. Kawarabayashi, S. Jegelka, Representation learning on graphs with jumping knowledge networks, in: Proceedings of the 35th International Conference on Machine Learning, Vol. 80, PMLR, 2018, pp. 5453–5462

    2018

  63. [72]

    H. Pei, B. Wei, K. C.-C. Chang, Y. Lei, B. Yang, Geom-gcn: Geometric graph convolutional networks, in: International Conference on Learning Representations, 2020

    2020

  64. [73]

    H. Zeng, H. Zhou, A. Srivastava, R. Kannan, V. Prasanna, Graphsaint: Graph sampling based inductive learning method, in: International Con- ference on Learning Representations, 2020

    2020

  65. [74]

    H. Zhao, L. Wei, Q. Yao, Simplifying architecture search for graph neural network, arXiv preprint arXiv:2008.11652 (2020). 43

    2020 arXiv

  66. [75]

    Fonseca, L

    C. Fonseca, L. Paquete, M. Lopez-Ibanez, An improved dimension-sweep algorithm for the hypervolume indicator, in: 2006 IEEE International Conference on Evolutionary Computation, IEEE, 2006, pp. 1157–1163. doi:10.1109/CEC.2006.1688440. 44

    2006 arXiv

Pith tools

Reviewed August 12, 2026 · model on record in the stance chip above.