Pith. sign in

REVIEW 2 major objections 1 minor 81 references

A meta-policy jointly selects ensemble surrogate configurations and infill criteria inside surrogate-assisted evolutionary algorithms.

Reviewed by Pith at T0; open to challenge. T0 means a machine referee read the full paper against a public rubric. the ladder, T0–T4 →

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

T0 review · grok-4.3

2026-06-28 17:39 UTC pith:XMHAFMUH

load-bearing objection AdaE-SAEA puts one RL meta-policy in charge of both ensemble surrogate choice and infill criterion to handle the robustness-accuracy trade-off, but the abstract supplies no experiments or ablations to check whether it actually works. the 2 major comments →

arxiv 2606.00862 v1 pith:XMHAFMUH submitted 2026-05-30 cs.NE cs.LG

Meta-Black-Box Optimization with Ensemble Surrogate Modeling for Robustness-Accuracy Trade-off within SAEA

Xiao Jin , Yongxiong Wang , Haobo Liu , Yudong Du , Yukun Du This is my paper
classification cs.NE cs.LG
keywords meta-black-box optimizationsurrogate-assisted evolutionary algorithmsensemble surrogate modelingreinforcement learningexpensive optimizationmulti-objective optimizationrobustness-accuracy trade-offinfill criterion
verification ladder T0 review T1 audit T2 compute T3 formal T4 reserved

The pith

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

The paper aims to remove the rigidity of hand-designed components in surrogate-assisted evolutionary algorithms by embedding them inside a meta-black-box optimization loop. A single learned policy decides both how to assemble an ensemble of surrogates—switching between bagging for early robustness and boosting for later accuracy—and which infill rule to apply at each step. Training uses reinforcement learning with parallel sampling and centralized updates so the same policy can transfer across tasks. Experiments on synthetic benchmarks and real-world expensive multi-objective problems show the resulting algorithm beats prior baselines and single-component meta methods. The work also checks that TabPFN works as a viable base model inside the ensemble.

Core claim

AdaE-SAEA places a standard SAEA inside the MetaBBO framework and trains one meta-policy to control both the choice of ensemble surrogate modules (bagging versus boosting) that trade robustness against accuracy across search phases and the choice of infill criterion that governs which new points are evaluated. The policy is trained by reinforcement learning that employs parallel sampling and centralized training for efficiency and transferability. On synthetic and real-world expensive multi-objective problems the resulting algorithm outperforms state-of-the-art fixed SAEAs and existing MetaBBO approaches while demonstrating that TabPFN can serve as an effective base surrogate.

What carries the argument

The meta-policy that simultaneously outputs both the ensemble configuration (bagging or boosting modules) for surrogate modeling and the infill criterion for adaptive sampling.

Load-bearing premise

A single meta-policy trained via RL with parallel sampling and centralized training can reliably select both infill criteria and ensemble configurations that balance robustness and accuracy across unseen tasks and search phases.

What would settle it

On a collection of new expensive multi-objective tasks the meta-policy fails to produce higher hypervolume or lower regret than fixed SAEA variants that use a single fixed ensemble and fixed infill rule.

Watch this falsifier — get emailed when new claim-graph text bears on it.

If this is right

  • The algorithm records higher performance than prior fixed SAEAs and single-component MetaBBO methods on both synthetic test functions and real-world expensive problems.
  • Ensemble modeling switches from robust early-stage behavior to accurate late-stage behavior without manual schedule design.
  • Parallel sampling plus centralized training yields a policy that transfers to new tasks without retraining from scratch.
  • TabPFN proves usable as the underlying base model inside the learned ensembles.
  • Unified control of surrogate construction and sampling is shown to be feasible inside the SAEA setting for the first time.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

  • The same joint-control pattern could be applied to other multi-component evolutionary or Bayesian optimizers that currently rely on separate hand-tuned modules.
  • Explicit robustness-accuracy scheduling inside ensembles may also improve surrogate-assisted methods on single-objective or noisy expensive problems.
  • The learned policy could be inspected to extract reusable heuristics about when bagging versus boosting is preferred at different stages of search.

Editorial analysis

A structured set of objections, weighed in public.

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

Referee Report

2 major / 1 minor

Summary. The paper proposes AdaE-SAEA, which embeds a surrogate-assisted evolutionary algorithm (SAEA) as the low-level optimizer in a MetaBBO framework. It jointly controls the infill criterion and ensemble surrogate modeling (bagging vs. boosting) via a single RL-trained meta-policy to balance robustness-accuracy trade-offs across search phases in expensive multi-objective optimization. The meta-policy uses parallel sampling and centralized training for efficiency and transferability. Experiments on synthetic and real-world problems claim outperformance over SOTA baselines and other MetaBBO methods; TabPFN is verified as a base surrogate. It positions itself as the first work to unify surrogate modeling and infill control while explicitly addressing the robustness-accuracy trade-off.

Significance. If the meta-policy reliably generalizes to unseen tasks and arbitrary search phases, the work would advance adaptive SAEAs by providing a unified, data-driven approach to component control that explicitly manages the robustness-accuracy trade-off. The RL training scheme with parallel sampling is a positive technical choice. However, the significance is constrained by the absence of evidence that performance gains reflect true transferability rather than task-specific fitting.

major comments (2)
  1. [Abstract] Abstract: the central claim that a single meta-policy 'can reliably select both infill criteria and ensemble configurations' on unseen tasks and phases is load-bearing, yet the abstract supplies no cross-task hold-out results, phase-specific ablations, or distribution-shift metrics. Outperformance on the reported suite could therefore reduce to task-specific tuning rather than the advertised unification.
  2. [Abstract] Abstract and experimental description: the assertion of outperformance over 'state-of-the-art baselines and MetaBBO-based methods' is presented without reference to statistical tests, ablation studies, or implementation details (e.g., how the meta-policy is validated for transfer). This directly undermines the claim that the unified control improves generalization.
minor comments (1)
  1. [Abstract] Abstract: the phrase 'to the best of our knowledge, this is the first work' should be supported by a brief comparison table or explicit citation of the closest prior MetaBBO-SAEA works to avoid overstatement.

Simulated Author's Rebuttal

2 responses · 0 unresolved

We thank the referee for the constructive feedback. We address the major comments point by point below, clarifying the evidence in the manuscript while agreeing to strengthen the abstract presentation.

read point-by-point responses

  1. Referee: [Abstract] Abstract: the central claim that a single meta-policy 'can reliably select both infill criteria and ensemble configurations' on unseen tasks and phases is load-bearing, yet the abstract supplies no cross-task hold-out results, phase-specific ablations, or distribution-shift metrics. Outperformance on the reported suite could therefore reduce to task-specific tuning rather than the advertised unification.

    Authors: The manuscript body details cross-task hold-out experiments on synthetic and real-world problems to assess meta-policy transferability, along with phase-specific analyses enabled by the parallel sampling and centralized training scheme. We agree the abstract's brevity omits explicit reference to these elements. We will revise the abstract to concisely note the hold-out validation and ablation support for the unified control. revision: yes

  2. Referee: [Abstract] Abstract and experimental description: the assertion of outperformance over 'state-of-the-art baselines and MetaBBO-based methods' is presented without reference to statistical tests, ablation studies, or implementation details (e.g., how the meta-policy is validated for transfer). This directly undermines the claim that the unified control improves generalization.

    Authors: The experimental section reports statistical tests (including Wilcoxon rank-sum) for comparisons against baselines and MetaBBO methods, plus ablations on ensemble strategies, infill criteria, and the RL training procedure for transfer. We will update the abstract to reference these supporting analyses for improved clarity on generalization claims. revision: yes

Circularity Check

0 steps flagged

No circularity detected; empirical algorithmic proposal with external experimental support.

full rationale

The provided abstract and context describe AdaE-SAEA as an RL-trained meta-policy that jointly selects ensemble surrogate configurations (bagging/boosting) and infill criteria within a MetaBBO framework for SAEAs. No equations, self-citations, or derivation steps are quoted that reduce any claimed prediction or result to its own fitted inputs by construction. Claims rest on outperformance versus baselines on synthetic and real-world problems plus TabPFN verification, which are independent empirical checks rather than self-referential. This matches the default case of a non-circular engineering paper.

Axiom & Free-Parameter Ledger

0 free parameters · 0 axioms · 0 invented entities

Only the abstract is available; the method description implies several free parameters typical of RL meta-optimization (policy network size, reward shaping, ensemble weights) and domain assumptions about surrogate error behavior, but none are enumerated in the provided text.

pith-pipeline@v0.9.1-grok · 5843 in / 1227 out tokens · 23769 ms · 2026-06-28T17:39:46.069215+00:00 · methodology

0 comments
read the original abstract

Surrogate-assisted evolutionary algorithms (SAEAs) have been widely used for expensive black-box optimization problems. However, their reliance on rigid and manually designed components limits their flexibility and generalization across tasks. Meta-black-box optimization (MetaBBO) provides a promising paradigm for adaptively configuring algorithmic components. Nevertheless, existing MetaBBO methods usually control only a single component, and few studies have investigated the unified control of multi-component optimizers such as SAEAs. Moreover, the robustness-accuracy trade-off in surrogate modeling, which is crucial for stable early-stage exploration and accurate late-stage exploitation, has rarely been explicitly considered. To address these issues, we propose AdaE-SAEA, an adaptive ensemble surrogate-assisted evolutionary algorithm for expensive multi-objective optimization. AdaE-SAEA embeds SAEA as the low-level optimizer within the MetaBBO framework and jointly controls the infill criterion and ensemble-based surrogate modeling. Specifically, bagging and boosting are designed as surrogate modeling modules to adaptively balance robustness and accuracy across different search phases, while the meta-policy simultaneously selects the infill criterion to enable adaptive sampling decisions. The meta-policy is trained through reinforcement learning with parallel sampling and centralized training, improving both training efficiency and transferability. Experiments on synthetic and real-world problems demonstrate that AdaE-SAEA outperforms state-of-the-art baselines and MetaBBO-based methods. We further verify the effectiveness of TabPFN as the base surrogate model for ensemble learning. To the best of our knowledge, this is the first work to unify the control of surrogate modeling and infill criteria in SAEAs while explicitly addressing the robustness--accuracy trade-off.

Figures

Figures reproduced from arXiv: 2606.00862 by Haobo Liu, Xiao Jin, Yongxiong Wang, Yudong Du, Yukun Du.

Figure 1
Figure 1. Figure 1: The general workflow of MetaBBO. is characterized using Exploratory Landscape Analysis (ELA) [12]. Once the configuration is applied, the lower-level optimizer solves the target prob￾lem and provides performance feedback to the meta controller. The goal of MetaBBO is to meta-learn a policy that maximizes cumulative performance gains over a distribution of tasks [13, 14]. By integrating this meta-level deci… view at source ↗
Figure 2
Figure 2. Figure 2: Bootstrap Sampling (p = 0.5) S1 S2 S3 ... ST Divide into Q Groups Data Subset M1 M2 M3 ... MT Trained Model Ensemble Averaging (Bagging Surrogate) Model Selection Train Base Model Mi Compute Residuals Train Base Model M(i+1) Update Residuals Repeat Weighted Combination of Models (Boosting Surrogate) Surrogate Prediction Bagging Boosting Early Stopping Historical Data Bootstrap Sampling (p = 0.5) S1 S2 S3 .… view at source ↗
Figure 3
Figure 3. Figure 3: The workflow of AdaE-SAEA. most promising candidates for expensive evaluations, and the newly evalu￾ated data are incorporated into the training set for the next iteration. The overall procedure is summarized in Algorithm 2 and illustrated in [PITH_FULL_IMAGE:figures/full_fig_p013_3.png] view at source ↗
Figure 4
Figure 4. Figure 4: Average reward per expensive evaluation. [PITH_FULL_IMAGE:figures/full_fig_p021_4.png] view at source ↗
Figure 5
Figure 5. Figure 5: Sensitivity analysis and ablation study of AdaE-SAEA using logarithmic HV ratio (base 2) 5.4. Sensitivity Analysis of nbs To examine the sensitivity of the proposed meta-policy to different values of nbs, we evaluate the performance of a meta-policy trained with nbs = 3 under alternative settings of nbs = 1, 5 and 10 during the testing phase. This analysis aims to investigate how varying the number of eval… view at source ↗
Figure 6
Figure 6. Figure 6: Comparison of surrogate modeling strategies As shown in the [PITH_FULL_IMAGE:figures/full_fig_p025_6.png] view at source ↗

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Reference graph

Works this paper leans on

81 extracted references · 7 canonical work pages

  1. [1]

    Google vizier: A service for black-box optimization

    Daniel Golovin, Benjamin Solnik, Subhodeep Moitra, Greg Kochanski, John Karro, and David Sculley. Google vizier: A service for black-box optimization. InProceedings of the 23rd ACM SIGKDD international conference on knowledge discovery and data mining, pages 1487–1495, 2017

    2017

  2. [2]

    Black-box optimization for automated discovery.Accounts of Chemical Research, 54(6):1334–1346, 2021

    Kei Terayama, Masato Sumita, Ryo Tamura, and Koji Tsuda. Black-box optimization for automated discovery.Accounts of Chemical Research, 54(6):1334–1346, 2021

    2021

  3. [3]

    Metabox: A benchmark platform for meta-black-box optimization with reinforcement learning

    Zeyuan Ma, Hongshu Guo, Jiacheng Chen, Zhenrui Li, Guojun Peng, Yue-Jiao Gong, Yining Ma, and Zhiguang Cao. Metabox: A benchmark platform for meta-black-box optimization with reinforcement learning. Advances in Neural Information Processing Systems, 36:10775–10795, 2023

    2023

  4. [4]

    Saea: A security-aware and energy-aware task scheduling strategy by parallel squirrel search algorithm in cloud envi- ronment.Expert Systems with Applications, 176:114915, 2021

    Behnam Mohammad Hasani Zade, Najme Mansouri, and Moham- mad Masoud Javidi. Saea: A security-aware and energy-aware task scheduling strategy by parallel squirrel search algorithm in cloud envi- ronment.Expert Systems with Applications, 176:114915, 2021

    2021

  5. [5]

    Sanjoy Chakraborty, Apu Kumar Saha, and Amit Chhabra. Improv- ing whale optimization algorithm with elite strategy and its application to engineering-design and cloud task scheduling problems.Cognitive Computation, 15(5):1497–1525, 2023

    2023

  6. [6]

    The rbf hyperparameter in evolutionary surrogate-assisted optimization

    Yoel Tenne. The rbf hyperparameter in evolutionary surrogate-assisted optimization. In2022 International Conference on Machine Learning, Control, and Robotics (MLCR), pages 38–42. IEEE, 2022

    2022

  7. [7]

    A comparative study on surrogate models for saeas.Optimization Letters, 14(8):2595–2614, 2020

    Mônica AC Valadão and Lucas S Batista. A comparative study on surrogate models for saeas.Optimization Letters, 14(8):2595–2614, 2020. 31

    2020

  8. [8]

    Decision space partition based surrogate-assisted evolutionary algorithm for expensive optimiza- tion.Expert Systems with Applications, 214:119075, 2023

    Yuanchao Liu, Jianchang Liu, and Shubin Tan. Decision space partition based surrogate-assisted evolutionary algorithm for expensive optimiza- tion.Expert Systems with Applications, 214:119075, 2023

    2023

  9. [9]

    Designx: Human-competitive algorithm designer for black-box optimization

    Hongshu Guo, Zeyuan Ma, Yining Ma, Xinglin Zhang, Wei-Neng Chen, and Yue-Jiao Gong. Designx: Human-competitive algorithm designer for black-box optimization. InThe Thirty-ninth Annual Conference on Neural Information Processing Systems

  10. [10]

    A surrogate-assisted evolution strategy for constrained multi-objective optimization.Expert Systems with Applications, 57:270–284, 2016

    Rituparna Datta and Rommel G Regis. A surrogate-assisted evolution strategy for constrained multi-objective optimization.Expert Systems with Applications, 57:270–284, 2016

    2016

  11. [11]

    A survey of surrogate-assisted evolutionary algorithms for expensive opti- mization.Journal of Membrane Computing, pages 1–20, 2024

    Jing Liang, Yahang Lou, Mingyuan Yu, Ying Bi, and Kunjie Yu. A survey of surrogate-assisted evolutionary algorithms for expensive opti- mization.Journal of Membrane Computing, pages 1–20, 2024

    2024

  12. [12]

    Auto-configuring exploration-exploitation tradeoff in evolution- ary computation via deep reinforcement learning

    Zeyuan Ma, Jiacheng Chen, Hongshu Guo, Yining Ma, and Yue-Jiao Gong. Auto-configuring exploration-exploitation tradeoff in evolution- ary computation via deep reinforcement learning. InProceedings of the Genetic and Evolutionary Computation Conference, pages 1497–1505, 2024

    2024

  13. [13]

    Multi-agent dynamic algorithm configuration

    Ke Xue, Jiacheng Xu, Lei Yuan, Miqing Li, Chao Qian, Zongzhang Zhang, and Yang Yu. Multi-agent dynamic algorithm configuration. Advances in Neural Information Processing Systems, 35:20147–20161, 2022

    2022

  14. [14]

    Symbol: Generating flexible black-box optimizers through symbolic equation learning.arXiv preprint arXiv:2402.02355, 2024

    Jiacheng Chen, Zeyuan Ma, Hongshu Guo, Yining Ma, Jie Zhang, and Yue-Jiao Gong. Symbol: Generating flexible black-box optimizers through symbolic equation learning.arXiv preprint arXiv:2402.02355, 2024

    work page arXiv 2024

  15. [15]

    Hongshu Guo, Yining Ma, Zeyuan Ma, Jiacheng Chen, Xinglin Zhang, Zhiguang Cao, Jun Zhang, and Yue-Jiao Gong. Deep reinforcement learning for dynamic algorithm selection: A proof-of-principle study on differential evolution.IEEE Transactions on Systems, Man, and Cyber- netics: Systems, 54(7):4247–4259, 2024

    2024

  16. [16]

    Deep reinforcement learning based parameter control in 32 differential evolution

    MuditaSharma, AlexandrosKomninos, ManuelLópez-Ibáñez, andDim- itar Kazakov. Deep reinforcement learning based parameter control in 32 differential evolution. InProceedings of the genetic and evolutionary computation conference, pages 709–717, 2019

    2019

  17. [17]

    A reinforcement learning level- based particle swarm optimization algorithm for large-scale optimiza- tion.Information Sciences, 602:298–312, 2022

    Feng Wang, Xujie Wang, and Shilei Sun. A reinforcement learning level- based particle swarm optimization algorithm for large-scale optimiza- tion.Information Sciences, 602:298–312, 2022

    2022

  18. [18]

    A hyperheuristic and rein- forcement learning guided meta-heuristic algorithm recommendation

    Ningning Zhu, Fuqing Zhao, and Jie Cao. A hyperheuristic and rein- forcement learning guided meta-heuristic algorithm recommendation. In 2024 27th International Conference on Computer Supported Cooperative Work in Design (CSCWD), pages 1061–1066. IEEE, 2024

    2024

  19. [19]

    Deep reinforcement learn- ing assisted surrogate model management for expensive constrained multi-objective optimization.Swarm and Evolutionary Computation, 92:101817, 2025

    Shuai Shao, Ye Tian, and Yajie Zhang. Deep reinforcement learn- ing assisted surrogate model management for expensive constrained multi-objective optimization.Swarm and Evolutionary Computation, 92:101817, 2025

    2025

  20. [20]

    An improved bagging ensemble surrogate-assisted evolutionary algorithm for expen- sive many-objective optimization.Applied Intelligence, 52(6):5949–5965, 2022

    Qinghua Gu, Xiaoyue Zhang, Lu Chen, and Naixue Xiong. An improved bagging ensemble surrogate-assisted evolutionary algorithm for expen- sive many-objective optimization.Applied Intelligence, 52(6):5949–5965, 2022

    2022

  21. [21]

    Offlinedata- driven evolutionary optimization using selective surrogate ensembles

    HandingWang, YaochuJin, ChaoliSun, andJohnDoherty. Offlinedata- driven evolutionary optimization using selective surrogate ensembles. IEEE Transactions on Evolutionary Computation, 23(2):203–216, 2018

    2018

  22. [22]

    A bagging-based surrogate-assisted evolutionary algorithm for expen- sive multi-objective optimization.Neural Computing and Applications, 34(14):12097–12118, 2022

    Yuanchao Liu, Jianchang Liu, Shubin Tan, Yongkuan Yang, and Fei Li. A bagging-based surrogate-assisted evolutionary algorithm for expen- sive multi-objective optimization.Neural Computing and Applications, 34(14):12097–12118, 2022

    2022

  23. [23]

    Zhenshou Song, Handing Wang, and Yaochu Jin. A surrogate-assisted evolutionary framework with regions of interests-based data selection for expensive constrained optimization.IEEE Transactions on Systems, Man, and Cybernetics: Systems, 53(10):6268–6280, 2023

    2023

  24. [24]

    The boosting approach to machine learning: An overview.Nonlinear estimation and classification, pages 149–171, 2003

    Robert E Schapire. The boosting approach to machine learning: An overview.Nonlinear estimation and classification, pages 149–171, 2003

    2003

  25. [25]

    Ray: A distributed framework for emerging {AI}applications

    Philipp Moritz, Robert Nishihara, Stephanie Wang, Alexey Tumanov, Richard Liaw, Eric Liang, Melih Elibol, Zongheng Yang, William Paul, 33 Michael I Jordan, et al. Ray: A distributed framework for emerging {AI}applications. In13th USENIX symposium on operating systems design and implementation (OSDI 18), pages 561–577, 2018

    2018

  26. [26]

    Accurate predictions on small data with a tabular foundation model.Nature, 637(8045):319–326, 2025

    Noah Hollmann, Samuel Müller, Lennart Purucker, Arjun Krishnaku- mar, Max Körfer, Shi Bin Hoo, Robin Tibor Schirrmeister, and Frank Hutter. Accurate predictions on small data with a tabular foundation model.Nature, 637(8045):319–326, 2025

    2025

  27. [27]

    Toward automated algorithm design: A survey and practical guide to meta-black-box-optimization.IEEE Transactions on Evolutionary Computation, 2025

    Zeyuan Ma, Hongshu Guo, Yue-Jiao Gong, Jun Zhang, and Kay Chen Tan. Toward automated algorithm design: A survey and practical guide to meta-black-box-optimization.IEEE Transactions on Evolutionary Computation, 2025

    2025

  28. [28]

    Meta-black- box optimization for evolutionary algorithms: Review and perspective

    Xu Yang, Rui Wang, Kaiwen Li, and Hisao Ishibuchi. Meta-black- box optimization for evolutionary algorithms: Review and perspective. Swarm and Evolutionary Computation, 93:101838, 2025

    2025

  29. [29]

    Meta-black-box optimization with bi-space landscape analysis and dual-control mechanism for saea, 2025

    Yukun Du, Haiyue Yu, Xiaotong Xie, Yan Zheng, Lixin Zhan, Yudong Du, Chongshuang Hu, Boxuan Wang, and Jiang Jiang. Meta-black-box optimization with bi-space landscape analysis and dual-control mecha- nism for saea.arXiv preprint arXiv:2511.15551, 2025

    work page arXiv 2025

  30. [30]

    Instance generation for meta-black-box optimization through latent space reverse engineering.arXiv preprint arXiv:2509.15810, 2025

    Chen Wang, Yue-Jiao Gong, Zhiguang Cao, and Zeyuan Ma. Instance generation for meta-black-box optimization through latent space reverse engineering.arXiv preprint arXiv:2509.15810, 2025

    work page arXiv 2025

  31. [31]

    Meta-black-box-optimization through offline q-function learning

    Zeyuan Ma, Zhiguang Cao, Zhou Jiang, Hongshu Guo, and Yue-Jiao Gong. Meta-black-box-optimization through offline q-function learning. arXiv preprint arXiv:2505.02010, 2025

    work page arXiv 2025

  32. [32]

    Metabox- v2: A unified benchmark platform for meta-black-box optimization

    Zeyuan Ma, Yue-Jiao Gong, Hongshu Guo, Wenjie Qiu, Sijie Ma, Hongqiao Lian, Jiajun Zhan, Kaixu Chen, Chen Wang, Zhiyang Huang, Zechuan Huang, Guojun Peng, Ran Cheng, and Yining Ma. Metabox- v2: A unified benchmark platform for meta-black-box optimization. In Advances in Neural Information Processing Systems, 2025

    2025

  33. [33]

    Automated selection of evolutionary multi-objective optimization algorithms

    Ye Tian, Shichen Peng, Tobias Rodemann, Xingyi Zhang, and Yaochu Jin. Automated selection of evolutionary multi-objective optimization algorithms. In2019 IEEE Symposium Series on Computational Intelli- gence (SSCI), pages 3225–3232. IEEE, 2019. 34

    2019

  34. [34]

    Selecting meta-heuristics for solv- ing vehicle routing problems with time windows via meta-learning.Ex- pert Systems with Applications, 118:470–481, 2019

    AndresEGutierrez-Rodríguez, SantiagoEConant-Pablos, JoséCOrtiz- Bayliss, and Hugo Terashima-Marín. Selecting meta-heuristics for solv- ing vehicle routing problems with time windows via meta-learning.Ex- pert Systems with Applications, 118:470–481, 2019

    2019

  35. [35]

    Large language model-enhanced algorithm selection: towards com- prehensive algorithm representation.arXiv preprint arXiv:2311.13184, 2023

    Xingyu Wu, Yan Zhong, Jibin Wu, Bingbing Jiang, and Kay Chen Tan. Large language model-enhanced algorithm selection: towards com- prehensive algorithm representation.arXiv preprint arXiv:2311.13184, 2023

    work page arXiv 2023

  36. [36]

    Reinforcement-learning-based parameter adap- tation method for particle swarm optimization.Complex & Intelligent Systems, 9(5):5585–5609, 2023

    Shiyuan Yin, Min Jin, Huaxiang Lu, Guoliang Gong, Wenyu Mao, Gang Chen, and Wenchang Li. Reinforcement-learning-based parameter adap- tation method for particle swarm optimization.Complex & Intelligent Systems, 9(5):5585–5609, 2023

    2023

  37. [37]

    Employing reinforcement learning to en- hance particle swarm optimization methods.Engineering Optimization, 54(2):329–348, 2022

    Di Wu and G Gary Wang. Employing reinforcement learning to en- hance particle swarm optimization methods.Engineering Optimization, 54(2):329–348, 2022

    2022

  38. [38]

    A new reinforcement learning-based memetic particle swarm optimizer.Applied Soft Computing, 43:276–297, 2016

    Hussein Samma, Chee Peng Lim, and Junita Mohamad Saleh. A new reinforcement learning-based memetic particle swarm optimizer.Applied Soft Computing, 43:276–297, 2016

    2016

  39. [39]

    Hongshu Guo, Zeyuan Ma, Jiacheng Chen, Yining Ma, Zhiguang Cao, Xinglin Zhang, and Yue-Jiao Gong. Configx: Modular configuration for evolutionary algorithms via multitask reinforcement learning.Proceed- ings of the AAAI Conference on Artificial Intelligence, pages 26982– 26990, 2025

    2025

  40. [40]

    Large language model for multiobjective evolutionary optimization

    Fei Liu, Xi Lin, Shunyu Yao, Zhenkun Wang, Xialiang Tong, Mingx- uan Yuan, and Qingfu Zhang. Large language model for multiobjective evolutionary optimization. InInternational Conference on Evolutionary Multi-Criterion Optimization, pages 178–191. Springer, 2025

    2025

  41. [41]

    Large language models as optimizers

    Chengrun Yang, Xuezhi Wang, Yifeng Lu, Hanxiao Liu, Quoc V Le, Denny Zhou, and Xinyun Chen. Large language models as optimizers. InThe Twelfth International Conference on Learning Representations, 2023

    2023

  42. [42]

    Automated design of metaheuristics using reinforcement learning within a novel general 35 search framework.IEEE Transactions on Evolutionary Computation, 27(4):1072–1084, 2022

    Wenjie Yi, Rong Qu, Licheng Jiao, and Ben Niu. Automated design of metaheuristics using reinforcement learning within a novel general 35 search framework.IEEE Transactions on Evolutionary Computation, 27(4):1072–1084, 2022

    2022

  43. [43]

    Neural exploratory landscape analysis for meta-black-box-optimization

    Zeyuan Ma, Jiacheng Chen, Hongshu Guo, and Yue-Jiao Gong. Neural exploratory landscape analysis for meta-black-box-optimization. InThe Thirteenth International Conference on Learning Representations, 2025

    2025

  44. [44]

    Llm driven design of continuous optimiza- tion problems with controllable high-level properties.arXiv preprint arXiv:2601.18846, 2026

    Urban Skvorc, Niki van Stein, Moritz Seiler, Britta Grimme, Thomas Bäck, and Heike Trautmann. Llm driven design of continuous optimiza- tion problems with controllable high-level properties.arXiv preprint arXiv:2601.18846, 2026

    work page arXiv 2026

  45. [45]

    Linear subspace surrogate modeling for large-scale expensive single/multi-objective optimization.IEEE Transactions on Evolutionary Computation, 2023

    Langchun Si, Xingyi Zhang, Ye Tian, Shangshang Yang, Limiao Zhang, and Yaochu Jin. Linear subspace surrogate modeling for large-scale expensive single/multi-objective optimization.IEEE Transactions on Evolutionary Computation, 2023

    2023

  46. [46]

    Safe: Scale-adaptive fitness evaluation method for expensive optimization problems.IEEE Transactions on Evolutionary Computation, 25(3):478–491, 2021

    Sheng-Hao Wu, Zhi-Hui Zhan, and Jun Zhang. Safe: Scale-adaptive fitness evaluation method for expensive optimization problems.IEEE Transactions on Evolutionary Computation, 25(3):478–491, 2021

    2021

  47. [47]

    Tao Zhang, Fuzhang Li, Xin Zhao, Wang Qi, and Tianwei Liu. A con- volutional neural network-based surrogate model for multi-objective op- timization evolutionary algorithm based on decomposition.Swarm and Evolutionary Computation, 72:101081, 2022

    2022

  48. [48]

    YajieZhang, HaoJiang, YeTian, HaipingMa, andXingyiZhang. Multi- granularitysurrogatemodelingforevolutionarymultiobjectiveoptimiza- tion with expensive constraints.IEEE Transactions on Neural Networks and Learning Systems, 35(3):2956–2968, 2023

    2023

  49. [49]

    A twofold infill criterion-driven heterogeneous ensemble surrogate-assisted evolutionary algorithm for computationally expensive problems.Knowledge-Based Systems, 236:107747, 2022

    Mingyuan Yu, Jing Liang, Zhou Wu, and Zhile Yang. A twofold infill criterion-driven heterogeneous ensemble surrogate-assisted evolutionary algorithm for computationally expensive problems.Knowledge-Based Systems, 236:107747, 2022

    2022

  50. [50]

    Heteroge- neous ensemble-based infill criterion for evolutionary multiobjective op- timization of expensive problems.IEEE transactions on cybernetics, 49(3):1012–1025, 2018

    Dan Guo, Yaochu Jin, Jinliang Ding, and Tianyou Chai. Heteroge- neous ensemble-based infill criterion for evolutionary multiobjective op- timization of expensive problems.IEEE transactions on cybernetics, 49(3):1012–1025, 2018. 36

    2018

  51. [51]

    Surrogate-assisted evolution- ary algorithm with hierarchical surrogate technique and adaptive infill strategy.Expert Systems with Applications, 232:120826, 2023

    Hao Chen, Weikun Li, and Weicheng Cui. Surrogate-assisted evolution- ary algorithm with hierarchical surrogate technique and adaptive infill strategy.Expert Systems with Applications, 232:120826, 2023

    2023

  52. [52]

    A review of surrogate-assisted evolutionary algorithms for expensive optimization problems.Expert Systems with Applications, 217:119495, 2023

    Chunlin He, Yong Zhang, Dunwei Gong, and Xinfang Ji. A review of surrogate-assisted evolutionary algorithms for expensive optimization problems.Expert Systems with Applications, 217:119495, 2023

    2023

  53. [53]

    A kriging-assisted two-archive evolutionary algorithm for expensive many- objective optimization.IEEE Transactions on Evolutionary Computa- tion, 25(6):1013–1027, 2021

    Zhenshou Song, Handing Wang, Cheng He, and Yaochu Jin. A kriging-assisted two-archive evolutionary algorithm for expensive many- objective optimization.IEEE Transactions on Evolutionary Computa- tion, 25(6):1013–1027, 2021

    2021

  54. [54]

    A surrogate-assisted constrained optimization evolutionary algorithm by searching multiple kinds of global and local regions.IEEE Transactions on Evolutionary Computation, 2023

    Yong Zeng, Yuansheng Cheng, and Jun Liu. A surrogate-assisted constrained optimization evolutionary algorithm by searching multiple kinds of global and local regions.IEEE Transactions on Evolutionary Computation, 2023

    2023

  55. [55]

    Kanak Kalita, Pradeep Jangir, Robert Čep, Sundaram B Pandya, and Laith Abualigah. Many-objective grasshopper optimization algorithm (maogoa): a new many-objective optimization technique for solving en- gineering design problems.International Journal of Computational In- telligence Systems, 17(1):214, 2024

    2024

  56. [56]

    A two- stage surrogate-assisted evolutionary algorithm (ts-saea) for expensive multi/many-objective optimization.Swarm and Evolutionary Compu- tation, 73:101107, 2022

    Jinglu Li, Peng Wang, Huachao Dong, and Jiangtao Shen. A two- stage surrogate-assisted evolutionary algorithm (ts-saea) for expensive multi/many-objective optimization.Swarm and Evolutionary Compu- tation, 73:101107, 2022

    2022

  57. [57]

    Classificationandregressiontrees, bagging, andboost- ing.Handbook of statistics, 24:303–329, 2005

    CliftonDSutton. Classificationandregressiontrees, bagging, andboost- ing.Handbook of statistics, 24:303–329, 2005

    2005

  58. [58]

    Gradient boosting machines, a tutorial

    Alexey Natekin and Alois Knoll. Gradient boosting machines, a tutorial. Frontiers in neurorobotics, 7:21, 2013

    2013

  59. [59]

    Amortized inference for causal structure learning.Ad- vances in Neural Information Processing Systems, 35:13104–13118, 2022

    Lars Lorch, Scott Sussex, Jonas Rothfuss, Andreas Krause, and Bern- hard Schölkopf. Amortized inference for causal structure learning.Ad- vances in Neural Information Processing Systems, 35:13104–13118, 2022

    2022

  60. [60]

    Tabpfn: A transformer that solves small tabular classification 37 problems in a second

    Noah Hollmann, Samuel Müller, Katharina Eggensperger, and Frank Hutter. Tabpfn: A transformer that solves small tabular classification 37 problems in a second. InThe Eleventh International Conference on Learning Representations, 2023

    2023

  61. [61]

    Transformers can do bayesian inference

    Samuel Müller, Noah Hollmann, Sebastian Pineda Arango, Josif Grabocka, and Frank Hutter. Transformers can do bayesian inference. InInternational Conference on Learning Representations, 2022

    2022

  62. [62]

    Kalyanmoy Deb and Himanshu Jain. An evolutionary many-objective optimization algorithm using reference-point-based nondominated sort- ing approach, part i: solving problems with box constraints.IEEE transactions on evolutionary computation, 18(4):577–601, 2013

    2013

  63. [63]

    Expensive multi-objective bayesian optimization based on diffusion models

    Bingdong Li, Zixiang Di, Yongfan Lu, Hong Qian, Feng Wang, Peng Yang, Ke Tang, and Aimin Zhou. Expensive multi-objective bayesian optimization based on diffusion models. InProceedings of the AAAI Conference on Artificial Intelligence, volume 39, pages 27063–27071, 2025

    2025

  64. [64]

    Differen- tiable expected hypervolume improvement for parallel multi-objective bayesian optimization

    Samuel Daulton, Maximilian Balandat, and Eytan Bakshy. Differen- tiable expected hypervolume improvement for parallel multi-objective bayesian optimization. In H. Larochelle, M. Ranzato, R. Hadsell, M.F. Balcan, and H. Lin, editors,Advances in Neural Information Processing Systems, volume 33, pages 9851–9864. Curran Associates, Inc., 2020

    2020

  65. [65]

    Data-driven evolution- ary sampling optimization forexpensive problems.Journal of Systems Engineering and Electronics, 32(2):318–330, 2021

    Zhen Huixiang, Gong Wenyin, and Wang Ling. Data-driven evolution- ary sampling optimization forexpensive problems.Journal of Systems Engineering and Electronics, 32(2):318–330, 2021

    2021

  66. [66]

    Yongcun Liu and Handing Wang. Surrogate-assisted hybrid evolution- ary algorithm with local estimation of distribution for expensive mixed- variable optimization problems.Applied Soft Computing, 133:109957, 2023

    2023

  67. [67]

    Multiobjective evolutionary algo- rithms: a comparative case study and the strength pareto approach

    Eckart Zitzler and Lothar Thiele. Multiobjective evolutionary algo- rithms: a comparative case study and the strength pareto approach. IEEE transactions on Evolutionary Computation, 3(4):257–271, 2002

    2002

  68. [68]

    A multi-stage expensive constrained multi-objective optimization algorithm based on ensemble infill crite- rion.IEEE Transactions on Evolutionary Computation, 2024

    Haofeng Wu, Qingda Chen, Jiaxin Chen, Yaochu Jin, Jinliang Ding, Xingyi Zhang, and Tianyou Chai. A multi-stage expensive constrained multi-objective optimization algorithm based on ensemble infill crite- rion.IEEE Transactions on Evolutionary Computation, 2024. 38

    2024

  69. [69]

    Exploratory landscape analysis

    Olaf Mersmann, Bernd Bischl, Heike Trautmann, Mike Preuss, Claus Weihs, and Günter Rudolph. Exploratory landscape analysis. InPro- ceedings of the 13th annual conference on Genetic and evolutionary com- putation, pages 829–836, 2011

    2011

  70. [70]

    Comprehensive feature-based landscape analysis of continuous and constrained optimization problems using the r-package flacco

    Pascal Kerschke and Heike Trautmann. Comprehensive feature-based landscape analysis of continuous and constrained optimization problems using the r-package flacco. InApplications in statistical computing: from music data analysis to industrial quality improvement, pages 93–123. Springer, 2019

    2019

  71. [71]

    Human-level control through deep reinforcement learning.nature, 518(7540):529–533, 2015

    Volodymyr Mnih, Koray Kavukcuoglu, David Silver, Andrei A Rusu, Joel Veness, Marc G Bellemare, Alex Graves, Martin Riedmiller, An- dreas K Fidjeland, Georg Ostrovski, et al. Human-level control through deep reinforcement learning.nature, 518(7540):529–533, 2015

    2015

  72. [72]

    Efficient multiobjective optimization employing gaussian processes, spectral sam- pling and a genetic algorithm.Journal of global optimization, 71(2):407– 438, 2018

    Eric Bradford, Artur M Schweidtmann, and Alexei Lapkin. Efficient multiobjective optimization employing gaussian processes, spectral sam- pling and a genetic algorithm.Journal of global optimization, 71(2):407– 438, 2018

    2018

  73. [73]

    Diversity-guided multi-objective bayesian optimization with batch eval- uations.Advances in neural information processing systems, 33:17708– 17720, 2020

    Mina Konakovic Lukovic, Yunsheng Tian, and Wojciech Matusik. Diversity-guided multi-objective bayesian optimization with batch eval- uations.Advances in neural information processing systems, 33:17708– 17720, 2020

    2020

  74. [74]

    Ex- pensive multiobjective optimization by moea/d with gaussian process model.IEEE Transactions on Evolutionary Computation, 14(3):456– 474, 2009

    Qingfu Zhang, Wudong Liu, Edward Tsang, and Botond Virginas. Ex- pensive multiobjective optimization by moea/d with gaussian process model.IEEE Transactions on Evolutionary Computation, 14(3):456– 474, 2009

    2009

  75. [75]

    Uncertainty-aware search framework for multi- objective bayesian optimization

    Syrine Belakaria, Aryan Deshwal, Nitthilan Kannappan Jayakodi, and Janardhan Rao Doppa. Uncertainty-aware search framework for multi- objective bayesian optimization. InProceedings of the AAAI Conference on Artificial Intelligence, volume 34, pages 10044–10052, 2020

    2020

  76. [76]

    A fast and elitist multiobjective genetic algorithm: Nsga-ii.IEEE transactions on evolutionary computation, 6(2):182–197, 2002

    Kalyanmoy Deb, Amrit Pratap, Sameer Agarwal, and TAMT Meyari- van. A fast and elitist multiobjective genetic algorithm: Nsga-ii.IEEE transactions on evolutionary computation, 6(2):182–197, 2002. 39

    2002

  77. [77]

    A fitness landscape ruggedness multiobjective differential evolution algorithm with a rein- forcement learning strategy.Applied Soft Computing, 96:106693, 2020

    Ying Huang, Wei Li, Furong Tian, and Xiang Meng. A fitness landscape ruggedness multiobjective differential evolution algorithm with a rein- forcement learning strategy.Applied Soft Computing, 96:106693, 2020

    2020

  78. [78]

    R2 indicator and deep reinforce- ment learning enhanced adaptive multi-objective evolutionary algo- rithm.arXiv preprint arXiv:2404.08161, 2024

    Farajollah Tahernezhad-Javazm, Debbie Rankin, Naomi Du Bois, Al- ice E Smith, and Damien Coyle. R2 indicator and deep reinforce- ment learning enhanced adaptive multi-objective evolutionary algo- rithm.arXiv preprint arXiv:2404.08161, 2024

    work page arXiv 2024

  79. [79]

    Comparison of multiobjective evolutionary algorithms: Empirical results.Evolution- ary computation, 8(2):173–195, 2000

    Eckart Zitzler, Kalyanmoy Deb, and Lothar Thiele. Comparison of multiobjective evolutionary algorithms: Empirical results.Evolution- ary computation, 8(2):173–195, 2000

    2000

  80. [80]

    Scalable test problems for evolutionary multiobjective optimization

    Kalyanmoy Deb, Lothar Thiele, Marco Laumanns, and Eckart Zitzler. Scalable test problems for evolutionary multiobjective optimization. In Evolutionary multiobjective optimization: theoretical advances and ap- plications, pages 105–145. Springer, 2005

    2005

Showing first 80 references.