Pith. sign in

REVIEW 5 major objections 5 minor 48 references

Data-Driven Antenna Miniaturization: A Knowledge-Based System Integrating Quantum PSO and Predictive Machine Learning Models

T0 review · 5 major / 5 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 A machine-learning-plus-quantum-PSO workflow designs a miniaturized loop-slot antenna in about 12 minutes, a 240x speedup over a 50-hour benchmark.

desk verdict A plausible antenna optimization workflow with one real dataset, but the 240x speedup claim ignores the cost of building the training set and the ML models miss the optimum by 9-18%. read the letter →

arxiv 2505.22440 v1 pith:GLJODY5D submitted 2025-05-28 cs.LG

Khan Masood Parvez , Sk Md Abidar Rahaman , Ali Shiri Sichani This is my paper
classification cs.LG
keywords AntennaMiniaturizationSlotQuantum-BehavedDynamicParticleSwarmOptimizationResonanceFrequencyPredictionMachineLearninginDesignAI-CADIntegrationSurrogateModeling
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

The paper is trying to establish that antenna miniaturization can be made a fast, almost fully automatic process instead of an expert-driven trial-and-error loop. The proposed system lets an optimizer choose the two loop diameters of a slot antenna, uses machine-learning models trained on 936 simulations to predict the resulting resonance frequency, and then checks the final geometry with one full-wave electromagnetic simulation. The authors report a resonance frequency of 1.4208 GHz, a 12.7 percent reduction from the conventional 1.60 GHz design, with the optimization and prediction steps taking about 12 seconds combined and the full cycle about 12.42 minutes on a standard desktop. They compare this with a 50-hour literature benchmark and claim a 240x acceleration. If the workflow holds up, it would turn a multi-day expert task into a same-session automated step for compact wireless antennas.

What carries the argument

The load-bearing mechanism is the loop-loaded slot antenna combined with the QDPSO update rule. Two inductive loops placed at the ends of the slot compensate the slot's capacitive reactance, so the structure resonates at a lower frequency; the difference between the outer and inner loop diameters controls the strength of that inductive effect. QDPSO updates a particle's position as $x_i(t+1)=p_i+\beta|p_i-g|\ln(1/u)$, where $p_i$ is the personal best, $g$ the global best, $\beta$ a contraction-expansion coefficient, and $u$ a uniform random number, and the fitness function $1/(1+|f_r-f_{\text{target}}|)$ rewards solutions that push the resonance frequency as far below the target as possible. The machine-learning layer, trained on 936 full-wave simulations, maps the two diameters to resonance frequency so that most electromagnetic solver calls can be skipped during optimization.

What would settle it

Re-run the complete workflow from scratch on the same desktop hardware, including all 936 training simulations, model training and hyperparameter tuning, and the final validation pass, and compare the total wall-clock time with 50 hours. If the total rivals or exceeds 50 hours, the claimed 240x acceleration collapses; given that the paper's single validation simulation took roughly 12 minutes, dataset generation alone could easily dominate the timeline.

Watch

Extended reading notes

Core claim

The central claim is that a quantum-behaved particle swarm optimizer can replace the manual search over loop dimensions and drive a loop-loaded slot antenna to a much lower resonance frequency while a machine-learning surrogate makes subsequent frequency estimates nearly free. For a fixed substrate, slot, and feed geometry, QDPSO varied only the inner and outer diameters of two identical loops and reached 1.4208 GHz, a 37.41 percent reduction from the 2.27 GHz reference slot and 12.7 percent below the earlier 1.60 GHz loop-loaded design. Among the four surrogate models, the stacked ensemble fit the training data best with an R2 of 0.9825, while SVM generalized best on the held-out validation set with an R2 of 0.7197. The authors state that the complete design cycle, including the full-wave electromagnetic validation, took 12.42 minutes on a standard desktop, and they present this as a 240x acceleration relative to the roughly 50 hours reported for the PSADEA method.

Load-bearing premise

The load-bearing premise is that the 12.42-minute design time is a fair comparison with the 50-hour PSADEA benchmark, which means the 936 full-wave simulations that build the training set and the time spent tuning the machine-learning models can be treated as one-time costs outside the reported cycle.

Editorial extensions

If this is right

  • If the reported 12.42-minute cycle is representative, a designer can move from a frequency target to a validated, fabrication-ready loop geometry within a single working session on commodity hardware.
  • The same QDPSO-plus-surrogate structure can be retargeted to other frequency bands or substrate properties by changing the fitness target and regenerating the training dataset for that fixed topology.
  • Because the surrogate predicts the resonance frequency in 0.75 seconds, the dominant remaining cost in the workflow is the one-time generation of training simulations, so further speedups would have to come from cheaper surrogates or reduced simulation counts.
  • The 12.7 percent resonance-frequency reduction (1.4208 GHz versus 1.60 GHz) gives a concrete, checkable miniaturization figure for compact IoT and consumer-radio antennas, assuming the simulated result survives physical measurement.

Reading between the lines

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

  • Editorial extension: the 240x speedup compares only the optimization-plus-prediction-plus-single-validation pass against the 50-hour benchmark; it excludes the 936 full-wave simulations needed to build the training set, so the true end-to-end time-to-first-design is likely days, not minutes.
  • Editorial extension: the study varies only two geometric parameters; expanding to more design variables (substrate size, slot dimensions, feed position) would require substantially more training data and is not supported by the reported accuracies.
  • Editorial extension: the surrogate models predict resonance frequency only, not return loss depth, bandwidth, or efficiency across the design space, so production-ready claims would need additional validated output channels.
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

5 major / 5 minor

Summary. The paper proposes a machine-learning-assisted antenna miniaturization workflow in which a Quantum-Behaved Dynamic Particle Swarm Optimization (QDPSO) algorithm searches for the inner and outer diameters of two loops loaded on a slot antenna, with the objective of lowering the resonance frequency. The authors train four ML models (SVM, Random Forest, XGBoost, and a stacked ensemble) on 936 HFSS simulation datasets to predict resonance frequency, and then validate the QDPSO-found optimum in ANSYS HFSS. The headline claims are a 12.7% resonance-frequency reduction (to 1.4208 GHz from a conventional 1.60 GHz design), a 12.42-minute 'complete design cycle' on a desktop PC, and a 240× speedup over a PSADEA-based benchmark. The paper also discusses a broader vision of automated antenna design for consumer wireless devices.

Significance. If the headline claims were reproducible and correctly computed, the proposed workflow would be of practical interest to antenna engineers seeking to replace manual trial-and-error loops with automated, ML-guided design. The paper does present a concrete antenna geometry, a 936-point simulation dataset, and a systematic comparison of four ML models with multiple error metrics, which is a useful starting point. However, the central quantitative contributions are not supported as reported: the miniaturization percentage is arithmetically incorrect, the speedup comparison omits the dominant offline cost of dataset generation, the optimization loop is driven by a closed-form approximation rather than HFSS simulation, and the ML predictions at the claimed optimum are inaccurate by 9–18%. These issues undermine the paper's main claims as they stand, although the underlying idea of combining an optimizer with a pre-trained surrogate is not without merit.

major comments (5)
  1. [Abstract and Section VII] The claimed '12.7 percent reduction' is arithmetically inconsistent with the stated frequencies. Using the quoted values, (1.60 − 1.4208)/1.60 = 11.2%, not 12.7%; the 12.7% figure appears to divide by the new frequency (1.4208 GHz) rather than the original 1.60 GHz. Since this percentage is a headline result, it should be corrected and re-verified.
  2. [Sections IV, VI, VII] The 'complete design cycle' of 12.42 minutes and the resulting '240× acceleration' exclude the time and computational cost of generating the 936 HFSS training simulations described in Section IV as synthesized using ANSYS Electronics Desktop. The paper's own single ANSYS validation run took 12 min 13.16 s; if a typical dataset simulation requires even a few minutes, building the dataset would take on the order of 190 hours, far exceeding the 50-hour PSADEA benchmark. The speedup claim is therefore not an end-to-end comparison and should be restated with the dataset-generation cost explicitly included.
  3. [Sections III-B and III-C] The fitness function used by QDPSO is computed from the closed-form approximations in Eqs. (5)–(6), not from full-wave HFSS simulation. The abstract's claim of a workflow 'integrating QDPSO with ANSYS HFSS simulations' is therefore misleading: HFSS is used to generate the training dataset and to validate the final design, but not inside the optimization loop. The 11.53-second optimization time is not comparable to full-wave simulation-driven methods such as PSADEA. The paper should clarify the actual coupling and assess the accuracy of Eq. (5) over the feasible parameter range.
  4. [Table VII and Section V] At the reported optimum (d_outer = 11.8614 mm, d_inner = 6.2441 mm), all four ML predictions (RF 1.5529, SVM 1.6576, XGB 1.6751, SM 1.6109 GHz) deviate from the ANSYS value of 1.4208 GHz by 0.13–0.25 GHz, i.e., 9.3–17.9%. This contradicts the text's statement that 'the predicted frequencies closely approximate the ANSYS results.' Given that the best validation R² is only 0.7197 (SVM) and the stacked model's validation R² is 0.4982, the ML models cannot substitute for HFSS validation, and the reported 0.75-second prediction time is not a practically useful substitute for the dominant simulation cost.
  5. [Sections VI–VII] The PSADEA baseline [49] addresses a different antenna problem (a slotted monopole for ultra-wideband body-centric imaging), so the comparison of '240× acceleration' is not controlled. The claimed speedup should be benchmarked against a full-wave optimization of the same loop-loaded slot antenna, or the statement should be downgraded to an illustrative order-of-magnitude comparison rather than a rigorous acceleration factor.
minor comments (5)
  1. [Section V] The figure references and captions are inconsistent: Fig. 10 is captioned 'Testing Phase' but the text describes it as validation data with R²=0.7197 (a value that appears in Table VI), and Fig. 11 is captioned 'Validation Phase' but the surrounding text repeats the same validation discussion. The captions and the text should be harmonized.
  2. [Abstract] The phrase 'This 240 times of acceleration' is grammatically awkward and should be revised to 'This 240× acceleration' or 'This 240-fold acceleration.'
  3. [Eq. (3)] The positional update in Eq. (3) uses only the positive branch of the QPSO update; standard QPSO uses a random choice of + or − for the term β·|p_i − g|·ln(1/u). The authors should specify the exact sign convention and report the values of the contraction–expansion coefficient β, the population size, and the number of iterations used in the experiments.
  4. [Section V] The stacking procedure as described uses base-model predictions on the training set to train the meta-model. Without out-of-fold predictions this risks information leakage and may explain the stacked model's poor validation performance. The paper should describe whether cross-validation was used for the stacking stage.
  5. [Section IV] The paper does not provide the dataset or the code, which limits reproducibility; consider making the 936-point HFSS dataset and the training scripts available as supplementary material.

Circularity Check

0 steps flagged · score 1.0 of 10

No circular derivation chain: the central numbers rest on HFSS simulation and a single HFSS validation, while the main weaknesses are omitted offline dataset cost and arithmetic inconsistency, not self-referential reasoning.

full rationale

The paper's derivation chain is not circular. The QDPSO search uses the closed-form resonance estimate in Eq. (5) inside the fitness function Eq. (4), but the reported resonance frequency of 1.4208 GHz is the ANSYS HFSS validation value shown in Fig. 5 and Table II, not the output of Eq. (5); optimizer guidance by an approximate formula and final full-wave validation are distinct steps. The ML models are trained on 936 independently synthesized HFSS data points and evaluated on held-out validation/test splits; Table VII predictions at the QDPSO optimum (e.g., SVM 1.6576 GHz vs. ANSYS 1.4208 GHz) deviate from the final validation value by about 9-18%, which quantitatively shows the predictions are not fitted to or forced to equal the claimed outcome. The 1.60 GHz baseline and 2.27 GHz reference frequency are taken from the authors' earlier published IEEE TAP paper [7]; this is a self-citation, but it is a published, externally checkable benchmark rather than an unverified premise, so it is not load-bearing circularity. The genuine problems are correctness and accounting issues rather than circularity: the 12.42-minute "complete design cycle" excludes the offline cost of generating the 936 HFSS training simulations, and (1.60 - 1.4208)/1.60 = 11.2%, not 12.7%. These flaws undermine the strength of the speedup and miniaturization claims but do not make any prediction equivalent to its inputs by construction. Score 1 reflects only the prominence of the self-cited baseline; no circular step was identified.

Assumptions & free parameters 4 free parameters · 5 assumptions · 0 invented entities

The central numerical claims depend on an unreported optimization configuration, a closed-form resonance formula that is not validated, the assumed representativeness of the 936-point dataset, and a speedup baseline of unknown comparability. No new physical entities are introduced.

free parameters (4)
  • QDPSO contraction-expansion coefficient beta
    Appears in Eq. (3); value not reported, so the 11.53-second optimization time cannot be reproduced.
  • QDPSO population size and iteration count
    Control parameters of the optimizer; not reported anywhere in the manuscript.
  • ML hyperparameters for SVM, Random Forest, XGBoost, and the stacked meta-model
    Not reported; Appendices A-C that should contain them are missing from the submitted text.
  • Train/validation/test split ratio = 90/5/5
    Chosen without justification; with 936 samples the validation and test sets contain only about 47 points each, which makes the reported R2 estimates unstable.
assumptions (5)
  • domain assumption The resonance frequency of the loop-loaded slot antenna is given by fr = c / (2 * (LSL + douter) * sqrt(eps_eff)) (Eq. 5).
    Introduced in Section III.B and used to define fitness; the paper never validates this closed-form model against the 936 HFSS simulations before relying on it for optimization.
  • domain assumption QDPSO with the update rule of Eq. (3) converges to the global optimum for this antenna problem.
    Taken from [19] without convergence analysis or seeds; no iteration traces other than a generic fitness curve are shown.
  • domain assumption The 936-point HFSS dataset is a sufficiently dense and accurate sample of the two-dimensional parameter space (dinner, douter).
    Used to train all ML models in Section IV; sampling grid, convergence criteria of HFSS, and dataset generation time are not reported.
  • ad hoc to paper The PSADEA benchmark [49] is a comparable baseline for the claimed 240x speedup.
    Section VI compares against a different slotted monopole for body-centric imaging without matching problem specification, hardware, or cost accounting.
  • domain assumption Machine learning models trained on the 936 HFSS runs generalize to unseen loop diameters, including the QDPSO optimum.
    The entire 'predictive' claim of Section V rests on this; validation R2 values (0.50 to 0.72) and Table VII errors of 9 to 18 percent at the optimum weaken this assumption.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Data-Driven Antenna Miniaturization: A Knowledge-Based System Integrating Quantum PSO and Predictive Machine Learning Models." pith.science (2026). https://pith.science/paper/GLJODY5D

@misc{pith2026250522440,
  author       = {Pith},
  title        = {Pith review of: Data-Driven Antenna Miniaturization: A Knowledge-Based System Integrating Quantum PSO and Predictive Machine Learning Models},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/GLJODY5D}},
  note         = {Machine review of arXiv:2505.22440}
}
read the original abstract

The rapid evolution of wireless technologies necessitates automated design frameworks to address antenna miniaturization and performance optimization within constrained development cycles. This study demonstrates a machine learning enhanced workflow integrating Quantum-Behaved Dynamic Particle Swarm Optimization (QDPSO) with ANSYS HFSS simulations to accelerate antenna design. The QDPSO algorithm autonomously optimized loop dimensions in 11.53 seconds, achieving a resonance frequency of 1.4208 GHz a 12.7 percent reduction compared to conventional 1.60 GHz designs. Machine learning models (SVM, Random Forest, XGBoost, and Stacked ensembles) predicted resonance frequencies in 0.75 seconds using 936 simulation datasets, with stacked models showing superior training accuracy (R2=0.9825) and SVM demonstrating optimal validation performance (R2=0.7197). The complete design cycle, encompassing optimization, prediction, and ANSYS validation, required 12.42 minutes on standard desktop hardware (Intel i5-8500, 16GB RAM), contrasting sharply with the 50-hour benchmark of PSADEA-based approaches. This 240 times of acceleration eliminates traditional trial-and-error methods that often extend beyond seven expert-led days. The system enables precise specifications of performance targets with automated generation of fabrication-ready parameters, particularly benefiting compact consumer devices requiring rapid frequency tuning. By bridging AI-driven optimization with CAD validation, this framework reduces engineering workloads while ensuring production-ready designs, establishing a scalable paradigm for next-generation RF systems in 6G and IoT applications.

Figures

Figures reproduced from arXiv: 2505.22440 by the authors.

Figure 1
Figure 1. 3D view of Slot antenna loaded with two loops. [PITH_FULL_IMAGE:figures/full_fig_p002_1.png] view at source ↗
Figure 3
Figure 3. Bottom surface of slot antenna with loop. [PITH_FULL_IMAGE:figures/full_fig_p002_3.png] view at source ↗
Figure 4
Figure 4. A Flowchart of the QDPSO setup.2-Faster Convergence: The quantum-inspired search in QDPSO accelerates convergence, identifying optimal an￾tenna designs more quickly than OptiSLang. 3-Enhanced Exploration: QDPSO’s quantum behavior improves solution space exploration, effectively avoiding local minima. 4- for Nonlinear Problems: QDPSO excels in optimizing complex, nonlinear antenna designs, outperforming some OptiSLan… view at source ↗
Figures from the paper (7 more)
Figure 5
Figure 5. Figure 5: Return Loss for loop loaded slot antenna with Outer Diameter of [PITH_FULL_IMAGE:figures/full_fig_p005_5.png]
Figure 6
Figure 6. Figure 6: Radiation pattern at frequency 1.4208 GHz [PITH_FULL_IMAGE:figures/full_fig_p005_6.png]
Figure 7
Figure 7. Figure 7: Steps to Predict the output [PITH_FULL_IMAGE:figures/full_fig_p005_7.png]
Figure 8
Figure 8. Figure 8: Fitness vs No. of Iterations machine learning algorithms, such as SVM [44], Random Forest [45], XG Boost[46], and Stacked Model [47] to predict resonance frequency using 936 datasets.The algorithms used to predict the resonance frequency are detailed in the appendices …
Figure 11
Figure 11. Figure 11: Scatter plots of the different models during the validation phase [PITH_FULL_IMAGE:figures/full_fig_p007_11.png]
Figure 9
Figure 9. Figure 9: Scatter plots of the different models during the Training phase [PITH_FULL_IMAGE:figures/full_fig_p007_9.png]
Figure 10
Figure 10. Figure 10: Scatter plots of the different models during the Testing phase [PITH_FULL_IMAGE:figures/full_fig_p007_10.png]

Discussion (0). Sign in to comment.

Reference graph

Works this paper leans on

48 extracted references · 48 canonical work pages

  1. [49]

    I. M. Danjuma, M. O. Akinsolu, C. H. See, R. A. Abd-Alhameed, and B. Liu, ”Design and optimization of a slotted monopole antenna for ultra- wide band body centric imaging applications,” IEEE J. Electromagn., RF Microw. Med. Biol., vol. 4, no. 2, pp. 140–147, Jun. 2020. PLACE PHOTO HERE Khan Masood Parvez (Student Member, IEEE)was born in Birbhum, India, i...

    work page 2020

  2. [1]

    Silver, Microwave Antenna Theory and Design , New York, NY , USA:McGraw-Hill, pp

    S. Silver, Microwave Antenna Theory and Design , New York, NY , USA:McGraw-Hill, pp. 296-298, 1946

    work page 1946

  3. [2]

    Fujimoto, H

    K. Fujimoto, H. Morishita, Modern Small Antennas . Cambridge, U.K.:Cambridge Univ. Press, 2013

    work page 2013

  4. [3]

    Design of an efficient miniaturized UHF planar antenna,

    K. Sarabandi and R. Azadegan, “Design of an efficient miniaturized UHF planar antenna,” IEEE Trans. Antennas Propag., vol. 51, no. 6, pp. 1270- 1276, June 2003

    work page 2003

  5. [4]

    A novel approach for miniaturization of slot antennas

    R. Azadegan and K. Sarabandi, “A novel approach for miniaturization of slot antennas”, IEEE Trans. Antennas Propag., vol. 51, no. 3, pp. 421-429, Mar. 2003

    work page 2003

  6. [5]

    A loop loading technique for the miniaturization of non-planar and planar antennas

    B. Ghosh, Sk. M. Haque, D. Mitra and S. Ghosh, “A loop loading technique for the miniaturization of non-planar and planar antennas”, IEEE Trans. Antennas Propag., vol. 58, no. 6, pp. 2116-2121, Jun. 2010

    work page 2010

  7. [6]

    Miniaturization of slot antennas using slit and strip loading

    B. Ghosh, Sk. M. Haque and D. Mitra, “Miniaturization of slot antennas using slit and strip loading”, IEEE Trans. Antennas Propag., vol. 59, no. 10, pp. 3922-3927, Oct. 2011

    work page 2011

  8. [7]

    Slot Antenna Miniaturization Us- ing Slit, Strip, and Loop Loading Techniques,

    SK. M. Haque and K. M. Parvez, “Slot Antenna Miniaturization Us- ing Slit, Strip, and Loop Loading Techniques,” IEEE Trans. Antennas Propag., vol. 65, no. 5, pp. 2215-2221, May 2017

    work page 2017

Show all 48 references
  1. [8]

    Miniaturized Wideband CP Antenna Using Hybrid Embedded Metasurface Structure,

    B. Zheng, N. Li, X. Li, X. Rao and Y . Shan, “Miniaturized Wideband CP Antenna Using Hybrid Embedded Metasurface Structure,” IEEE Access, vol. 10, pp. 120056-120062, 2022

    2022

  2. [9]

    A Miniaturized Differentially Fed Patch Antenna Based on Capacitive Slots,

    L. Wang, X. -X. Yang, T. Lou and S. Gao, “A Miniaturized Differentially Fed Patch Antenna Based on Capacitive Slots,” IEEE Antennas Wirel. Propag. Lett., vol. 21, no. 7, pp. 1472-1476, July 2022

    2022

  3. [10]

    Design of Miniaturized High Gain Bow-Tie Antenna,

    Y . Li and J. Chen, “Design of Miniaturized High Gain Bow-Tie Antenna,” IEEE Trans. Antennas Propag ., vol. 70, no. 1, pp. 738-743, Jan. 2022

    2022

  4. [11]

    3-D-Corrugated Ground Structure: A Microstrip Antenna Miniaturization Technique,

    P. B. Samal, S. J. Chen and C. Fumeaux, “3-D-Corrugated Ground Structure: A Microstrip Antenna Miniaturization Technique,”IEEE Trans. Antennas Propag., vol. 72, no. 5, pp. 4010-4022,May 2024

    2024

  5. [12]

    A Study of Wideband and Compact Slot Antennas Utilizing Special Dispersive Materials,

    X. Yang, I. Calisir, L. Chen, E. L. Bennett, J. Xiao and Y . Huang, “A Study of Wideband and Compact Slot Antennas Utilizing Special Dispersive Materials,” IEEE Open J. Antennas Propag ,vol. 5, no. 5, pp. 1414-1422, Oct. 2024

    2024

  6. [13]

    Broadband RCS Reduction, Antenna Miniaturization, and Bandwidth Enhancement by Combining Reactive Impedance Surface and Polariza- tion Conversion Metasurface,

    Q. Zheng, W. Liu, Q. Zhao, L. Kong, Y . -H. Ren and X. -X. Yang, “Broadband RCS Reduction, Antenna Miniaturization, and Bandwidth Enhancement by Combining Reactive Impedance Surface and Polariza- tion Conversion Metasurface,” IEEE Trans. Antennas Propag. , vol. 72, no. 9, pp. ...

    2024

  7. [14]

    Design of a Biomedical Antenna System for Wireless Communication of Ingestible Capsule Endoscope,

    R. Gao, H. Sun, R. Ren and H. Zhang, “Design of a Biomedical Antenna System for Wireless Communication of Ingestible Capsule Endoscope,” IEEE Antennas Wirel. Propag. Lett. , vol. 23, no. 12, pp.4243-4247, Dec. 2024

    2024

  8. [15]

    A Compact Self-Triplexing Antenna for Head Implants,

    A. Iqbal, M. Al-Hasan, I. Ben Mabrouk and T. A. Denidni, “A Compact Self-Triplexing Antenna for Head Implants,” IEEE Trans. Antennas Propag., vol. 72, no. 11, pp. 8207-8214, Nov. 2024

    2024

  9. [16]

    The explosion of artificial intelligence in antennas and propagation: How deep learning is advancing our state of the art,

    S. D. Campbell, R. P. Jenkins, P. J. O’Connor, and D. Werner, “The explosion of artificial intelligence in antennas and propagation: How deep learning is advancing our state of the art,” IEEE Antennas Propag. Mag., vol. 63, no. 3, pp. 16–27, Jun. 2021

    2021

  10. [17]

    Antennas and propagation research from large-scale unstructured data with machine learning: A review and predictions,

    Y . O. Cha, A. A. Ihalage, and Y . Hao, “Antennas and propagation research from large-scale unstructured data with machine learning: A review and predictions,” IEEE Antennas Propag. Mag. , vol. 65,no. 5, pp. 10–24, Oct. 2023

    2023

  11. [18]

    Particle swarm optimization in electromagnetics,

    J. Robinson and Y . Rahmat-Samii, “Particle swarm optimization in electromagnetics,” IEEE Trans. Antennas Propag. , vol. 52, no. 2,pp. 397–407, Feb. 2004

    2004

  12. [19]

    Quantum Particle Swarm Optimization for Electromagnetics,

    S. M. Mikki and A. A. Kishk, “Quantum Particle Swarm Optimization for Electromagnetics,” IEEE Trans. Antennas Propag. , vol. 54, no. 10, pp. 2764-2775, Oct. 2006

    2006

  13. [20]

    A hybrid smart quantum particle swarm optimization for multimodal electromag- netic design problems,

    S. Ahad, S. Yang, S. U. Khan, S. A. Khan, and R. A. Khan, “A hybrid smart quantum particle swarm optimization for multimodal electromag- netic design problems,” IEEE Access, vol. 10, pp. 72339–72347, 2022

    2022

  14. [21]

    Multi-modal multi-objective particle swarm optimization with self-adjusting strategy,

    H. Han, Y . Liu, Y . Hou, and J. Qiao, “Multi-modal multi-objective particle swarm optimization with self-adjusting strategy,” Information Sciences, vol. 629, pp. 580–598, 2023

    2023

  15. [22]

    A hybrid Taguchi binary particle swarm optimization for antenna designs,

    X. Jia and G. Lu, “A hybrid Taguchi binary particle swarm optimization for antenna designs,” IEEE Antennas Wireless Propag. Lett. , vol. 18, no. 8, pp. 1581–1585, Aug. 2019

    2019

  16. [23]

    Beamsteering and Beamshaping Using a Linear Antenna Array Based on Particle Swarm Optimization

    Lukasz A. Greda, Andreas Winterstein, Daniel L. Lemes, Marcos V . T. Heckler, “Beamsteering and Beamshaping Using a Linear Antenna Array Based on Particle Swarm Optimization”, IEEE Access, vol.7, pp.141562- 141573, 2019

    2019

  17. [24]

    Machine-Learning- Assisted Optimization for Antenna Geometry Design,

    Q. Wu, W. Chen, C. Yu, H. Wang and W. Hong, “Machine-Learning- Assisted Optimization for Antenna Geometry Design,” IEEE Trans. Antennas Propag., vol. 72, no. 3, pp. 2083-2095, March 2024

    2024

  18. [25]

    Multilayer machine learning-assisted optimization-based robust design and its applications to antennas and array,

    Q. Wu, W. Chen, C. Yu, H. Wang, and W. Hong, “Multilayer machine learning-assisted optimization-based robust design and its applications to antennas and array,” IEEE Trans. Antennas Propag. , vol. 69, no. 9,pp. 6052–6057, Sep. 2021

    2021

  19. [26]

    Multibranch Machine Learning-Assisted Optimization and Its Application to Antenna Design,

    W. Chen, Q. Wu, C. Yu, H. Wang and W. Hong, “Multibranch Machine Learning-Assisted Optimization and Its Application to Antenna Design,” IEEE Trans. Antennas Propag., vol. 70, no. 7, pp. 4985-4996, July 2022

    2022

  20. [28]

    An efficient method for complex antenna design based on a self adaptive surrogate model-assisted optimization technique,

    B. Liu et al., “An efficient method for complex antenna design based on a self adaptive surrogate model-assisted optimization technique,” IEEE Trans. Antennas Propag., vol. 69, no. 4, pp. 2302–2315, Apr. 2021

    2021

  21. [29]

    A trust-region parallel Bayesian optimization method for simulation-driven antenna design,

    J. Zhou et al., “A trust-region parallel Bayesian optimization method for simulation-driven antenna design,” IEEE Trans. Antennas Propag. , vol. 69, no. 7, pp. 3966–3981, Jul. 2021

    2021

  22. [30]

    An efficient method for antenna design based on a selfadaptive Bayesian neural network assisted global optimization tech- nique,

    Y . Liu et al., “An efficient method for antenna design based on a selfadaptive Bayesian neural network assisted global optimization tech- nique,” IEEE Trans. Antennas Propag., vol. 70, no. 12, pp. 11375–11388, Dec.2022

    2022

  23. [31]

    Tolerance optimization of antenna structures by means of response feature surrogates,

    S. Koziel and A. Pietrenko-Dabrowska, “Tolerance optimization of antenna structures by means of response feature surrogates,” IEEE Trans. Antennas Propag., vol. 70, no. 11, pp. 10988–10997, Nov. 2022

    2022

  24. [32]

    Generalized formulation of re- sponse features for reliable optimization of antenna input characteristics,

    A. Pietrenko-Dabrowska and S. Koziel, “Generalized formulation of re- sponse features for reliable optimization of antenna input characteristics,” IEEE Trans. Antennas Propag., vol. 70, no. 5, pp. 3733–3748, May 2022

    2022

  25. [33]

    Machine learning techniques for optimizing design of double T-shaped monopole antenna,

    Y . Sharma, H. H. Zhang, and H. Xin, “Machine learning techniques for optimizing design of double T-shaped monopole antenna,” IEEE Trans.Antennas Propag., vol. 68, no. 7, pp. 5658–5663, Jul. 2020

    2020

  26. [34]

    Ma- chine learning methods-based modeling and optimization of 3-Dprinted dielectrics around monopole antenna,

    Y . Sharma, X. Chen, J. Wu, Q. Zhou, H. H. Zhang, and H. Xin, “Ma- chine learning methods-based modeling and optimization of 3-Dprinted dielectrics around monopole antenna,” IEEE Trans. Antennas Propag. , vol. 70, no. 7, pp. 4997–5006, Jul. 2022

    2022

  27. [35]

    Beam forming design for large-scale antenna arrays using deep learning,

    T. Lin and Y . Zhu, “Beam forming design for large-scale antenna arrays using deep learning,” IEEE Wireless Commun. Lett. , vol. 9, no. 1,pp. 103–107, Jan. 2020

    2020

  28. [36]

    Deep learning based antenna design and beam-steering capabilities for millimeter-wave applications,

    A. M. Montaser and K. R. Mahmoud, “Deep learning based antenna design and beam-steering capabilities for millimeter-wave applications,” IEEE Access, vol. 9, pp. 145583–145591, 2021

    2021

  29. [37]

    Multilayer machine learning-assisted optimization-based robust design and its applications to antennas and array,

    Q. Wu, W. Chen, C. Yu, H. Wang, and W. Hong, “Multilayer machine learning-assisted optimization-based robust design and its applications to antennas and array,” IEEE Trans. Antennas Propag. , vol. 69, no. 9,pp.6052–6057, Sep. 2021

    2021

  30. [38]

    On formulating and de- signing antenna arrays by evolutionary algorithms,

    Q. Xu, S. Zeng, F. Zhao, R. Jiao, and C. Li, “On formulating and de- signing antenna arrays by evolutionary algorithms,” IEEE Trans.Antennas Propag., vol. 69, no. 2, pp. 1118–1129, Feb. 2021

    2021

  31. [39]

    Thin-wire antenna design using a novel branching scheme and genetic algorithm optimization,

    J. S. Smith and M. E. Baginski, “Thin-wire antenna design using a novel branching scheme and genetic algorithm optimization,”IEEE Trans. Antennas Propag., vol. 67, no. 5, pp. 2934–2941, May 2019

    2019

  32. [40]

    Herscovici, M

    N. Herscovici, M. F. Osorio and C. Peixeiro, ”Miniaturization of rectangular microstrip patches using genetic algorithms,” IEEE Antennas Wirel. Propag. Lett., vol. 1, pp. 94-97, 2002

    2002

  33. [41]

    Wideband printed monopole design using genetic algorithm,

    John, M. and M. J. Ammann, “Wideband printed monopole design using genetic algorithm,” IEEE Antennas Wirel. Propag. Lett., V ol. 6, 447–449, 2007

    2007

  34. [42]

    An intelligent antenna synthesis method based on machine learning,

    D. Shi, C. Lian, K. Cui, Y . Chen, and X. Liu, “An intelligent antenna synthesis method based on machine learning,” IEEE Trans. Antennas Propag., vol. 70, no. 7, pp. 4965–4976, Jul. 2022

    2022

  35. [43]

    ANSYS Corporation Ver 18.2 ,“ HFSS,” Pittsburgh, PA, 15317

  36. [44]

    Im- provements to platt’s SMO algorithm for SVM classifier design,

    Keerthi SS, Shevade SK, Bhattacharyya C, Radha Krishna MK, “Im- provements to platt’s SMO algorithm for SVM classifier design,” Neural Comput., 2001;13(3):637–49

    2001

  37. [45]

    Random forests

    Breiman L. Random forests. Mach Learn. 2001;45(1):5–32. JOURNAL OF LATEX CLASS FILES, VOL. XX, NO. XX, XX 20XX 10

    2001

  38. [46]

    Scikit-learn: machine learning in python

    Pedregosa F, Varoquaux G, Gramfort A, Michel V , Thirion B, Grisel O, Blondel M, Prettenhofer P, Weiss R, Dubourg V , et al. Scikit-learn: machine learning in python. J Mach Learn Res. 2011;12:2825–30

    2011

  39. [47]

    Stacked generalization,

    D. H. Wolpert, “Stacked generalization,” Neural Networks , vol. 5, no. 2, pp. 241-259, 1992

    1992

  40. [48]

    Introduction to VLSI Systems,

    C. Mead and L. Conway, “Introduction to VLSI Systems,” IEEE Journal of Solid-State Circuits , vol. 15, no. 6, pp. 1068-1075, Dec. 1980

    1980

Pith tools

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