Lightweight Convolutional Neural Networks for Retinal Disease Classification

Abdul Jalil M. Khalaf; Abir Jaafar Hussain; Duaa Kareem Qasim; Lafta Raheem Ali; Sabah Abdulazeez Jebur

REVIEW 4 major objections 5 minor 24 references

Lightweight Convolutional Neural Networks for Retinal Disease Classification

T0 review · 4 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 This paper claims that two lightweight pretrained CNNs, MobileNetV2 and NASNetMobile, classify normal, diabetic-retinopathy, and macular-hole fundus images with 90.8% and 89.5% accuracy, respectively, on a three-class subset of the RFMiD…

desk verdict Routine transfer-learning benchmark whose headline accuracy is plausible but rests on an underspecified dataset subset and a single split; the 90.8% vs 89.5% gap is within noise. read the letter →

arxiv 2506.03186 v1 pith:SS2RRSAB submitted 2025-05-30 eess.IV cs.AIcs.CVcs.LGcs.NE

Duaa Kareem Qasim , Sabah Abdulazeez Jebur , Lafta Raheem Ali , Abdul Jalil M. Khalaf , Abir Jaafar Hussain This is my paper

classification eess.IVcs.AIcs.CVcs.LGcs.NE

keywords MobileNetV2NASNetretinaldiseaseclassificationfundusimagesRFMiDdatasettransferlearningdiabeticretinopathymacularhole

verification ladder T0 review T1 audit T2 compute T3 formal

The pith

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

The reading

This paper seeks to establish that off-the-shelf lightweight convolutional networks can serve as a practical screening tool for two sight-threatening retinal diseases. It trains two ImageNet-pretrained mobile architectures, MobileNetV2 and NASNetMobile, on a three-class subset of the RFMiD fundus-image dataset containing normal retinas, diabetic retinopathy, and macular hole. The claimed result is that MobileNetV2 reaches 90.8% accuracy and NASNetMobile 89.5%, with recall, precision, and F1-score all near 90%. If true, this means high-accuracy retinal triage does not require heavy, compute-intensive models, which matters for deployment in low-resource clinics.

What carries the argument

The argument is carried by two pretrained lightweight CNN architectures, MobileNetV2 and NASNetMobile, adapted by removing their original fully connected heads and adding a Global Average Pooling layer, a Flatten layer, a 1024-unit Dense layer with ReLU, a 30% dropout, and a three-way Softmax. MobileNetV2's depthwise-separable bottleneck blocks and NASNetMobile's Normal and Reduction Cells with depthwise-separable convolutions are what keep parameter counts low, while transfer learning from ImageNet supplies the visual features. Preprocessing, consisting of resizing to 224x224, min-max normalization, and augmentation with rotation, shifts, shear, zoom, and horizontal flip, is what lets the small dataset support 30 epochs of fine-tuning with the Adam optimizer.

What would settle it

Compute the same MobileNetV2 and NASNetMobile accuracies on a version of the RFMiD three-class subset that removes images with multiple simultaneous diseases and splits by patient ID rather than by image; if accuracy falls well below 90%, the claimed screening-level performance does not survive.

Watch

Extended reading notes

Core claim

The central claim is that a fine-tuned MobileNetV2 classifies a three-class subset of RFMiD with 90.8% accuracy, edging out NASNetMobile at 89.5% and beating published baselines including EfficientNetB3 (90%), a 20-layer CNN (90.34%), and Xception (90.78%) on the same dataset. The paper also claims both lightweight models keep recall, precision, and F1-scores between 89% and 91%, indicating that no single class is being sacrificed for overall accuracy. The discovery, on the paper's own terms, is that transfer-learned mobile CNNs give near-90% accuracy with far lower computational cost than larger models, providing a foundation for AI-assisted ophthalmic diagnosis.

Load-bearing premise

The reported accuracies are trustworthy only if the RFMiD ground-truth labels are correct and if the random three-class split keeps every patient's images entirely in one partition without leaking multi-disease examples.

Editorial extensions

If this is right

If MobileNetV2's 90.8% accuracy holds, a lightweight model can serve as a screening triage for normal, diabetic-retinopathy, and macular-hole fundus images on modest hardware.
Both models' balanced recall and precision around 90% imply that neither diabetic retinopathy nor macular hole is systematically missed in this three-class setup.
The comparison results imply that resource-efficient models can match or exceed heavier published methods on the RFMiD three-class task.
The method's train/test split and augmentation protocol, as described, would form the basis for extending the approach to more disease classes and larger datasets.

Reading between the lines

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

A natural test the paper leaves implicit is splitting RFMiD by patient rather than by image; if images from the same patient leak across train and test partitions, the reported accuracies could overstate real-world screening performance.
The paper does not report class-wise sensitivity, so a follow-up analysis should check whether DR and MH recall differ sharply; if they do, the overall accuracy could hide a screening failure for one disease.
Because RFMiD is originally a 46-class multi-label dataset, a cleaned re-derivation of the three-class subset that specifies how multi-disease images were excluded would let others independently verify the 90.8% figure.
Since both models are pretrained on ImageNet, the method should plausibly transfer to other fundus cameras and disease taxonomies, but that transferability remains untested here.

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.

Desk Editor's Note

read the letter

You should know two things. First, this is a standard fine-tuning benchmark: MobileNetV2 and NASNetMobile pretrained on ImageNet, retrained on a three-class subset of RFMiD (Normal/DR/MH), with data augmentation. No new method, no new mechanism, no new dataset. Second, the central claim — 90.8% vs 89.5% — is plausible but not well supported, for reasons that are fixable.

What the paper does well: the writing is clear, the choice of lightweight models is sensible for a deployment story, and the authors report accuracy, recall, precision, F1, and confusion matrices. Using RFMiD is a plus because it's public. The related work is adequate, though it already contains many similar CNN-plus-transfer-learning studies.

The big soft spot is the dataset description. The abstract says 3,200 fundus images, but Table I totals 1,451. The paper never says how the three-class subset was carved from the 46-class multi-label RFMiD. Were images with co-occurring pathologies discarded or assigned one label? Was the train/test split done per patient or per image? This matters: if same-patient images leaked across the split, accuracy is inflated; if multi-disease images remain, the labels are ambiguous. The stress-test note is right about this. Also, there is a single random split with no confidence intervals, no repeated runs, no statistical test. With 362 test images, the 1.3-point gap between MobileNetV2 and NASNetMobile is roughly half a standard error, so 'outperforming' is overclaiming. Table III compares against methods with different class sets and protocols; the claim that MobileNetV2 'outperformed all compared models' is not backed by a controlled comparison. No code or seeds are provided, so independent verification is impossible.

I don't think there's anything dishonest here; it's a reporting problem. The methodology is standard and the numbers are the kind you'd expect from fine-tuning on a clean subset. But as written, the headline number is not a trustworthy benchmark.

For a reader collecting baselines for lightweight models on fundus images, this is a data point to be aware of, not a result to build on. It deserves a referee only if the authors can fix the reporting: specify subset exclusion rules, do patient-level stratified splits, run repeated trials with confidence intervals, and release code. I'd send it to review with that condition, but I wouldn't lose sleep if it were desk rejected on novelty. My recommendation: treat as a conditional accept for review, not a reject for fraud.

Referee Report

4 major / 5 minor

Summary. The paper fine-tunes two ImageNet-pretrained lightweight CNNs, MobileNetV2 and NASNetMobile, on a three-class subset (Normal, Diabetic Retinopathy, Macular Hole) of the RFMiD fundus image dataset and reports classification performance. The authors report that MobileNetV2 achieves 90.8% accuracy, NASNetMobile 89.5% accuracy, and that MobileNetV2 outperforms all compared methods in a literature comparison. The stated contribution is a demonstration that lightweight transfer-learned CNNs can classify these retinal conditions with high accuracy while remaining computationally efficient.

Significance. If the central accuracy estimate is valid, the paper provides a useful but modest data point: two off-the-shelf lightweight architectures, with transfer learning and augmentation, reach roughly 90% accuracy on a three-class subset of a public fundus dataset. The work is empirical and externally validated on a held-out test set, so there is no circularity in the main claim. However, the paper's strengths are undercut by insufficient description of the dataset subsetting procedure, a single unstratified random split without uncertainty quantification, and a comparison table that mixes different protocols. The contribution is appropriate for a workshop-level or applied journal venue, but the current manuscript does not yet support its headline claims as rigorously as needed.

major comments (4)

[Section III-A and Table I] The dataset description is internally inconsistent: the text states that RFMiD consists of 3,200 fundus images, but Table I lists only 401+376+312 = 1,089 training images and 134+124+104 = 362 test images, totaling 1,451 images. The paper never explains how the three-class subset was carved out of the 46-class multi-label dataset, including whether images containing more than one disease were discarded, assigned a single label, or kept with overlapping labels, and how 'Normal' was defined. This is load-bearing because label ambiguity directly affects the measured accuracy; without an explicit exclusion rule, the reported 90.8% figure cannot be interpreted or reproduced.
[Section III-A and Section IV-B] The evaluation rests on a single randomized split with no patient-level stratification, no repeated runs, and no confidence intervals or significance tests. With 362 test images, the 1.3% accuracy difference between MobileNetV2 (90.8%) and NASNetMobile (89.5%) is within the margin of error; the claim that MobileNetV2 'outperformed' NASNetMobile is therefore not statistically supported. The authors should report multiple seeds, confidence intervals, or at least a McNemar test, and should clarify whether the split was performed at the patient level to avoid leakage from same-patient images appearing in both training and test sets.
[Section IV-C and Table III] The comparison in Table III is not apples-to-apples. The cited methods use different class sets (e.g., multi-label classification, binary classification, different disease subsets), different dataset splits, and possibly different numbers of classes; a direct comparison of raw accuracy numbers under these conditions is not meaningful. The sentence 'MobileNetV2 outperformed all compared models' is therefore unsupported by the evidence presented. The authors should either restrict the comparison to methods evaluated on the same three-class protocol or explicitly state the protocol differences and refrain from claiming superiority.
[Section IV-B and Figure 3] The paper reports only macro-level accuracy, recall, precision, and F1-score, without per-class metrics or the numeric confusion matrix. Given the class sizes in Table I (e.g., MH has only 312 training images), per-class performance is essential to judge whether the model is clinically plausible or merely exploiting class imbalance. Figure 3 shows confusion matrices and curves, but the subfigures are not labeled or described in enough detail to extract the per-class numbers; the authors should provide a numeric confusion matrix and per-class precision/recall for both models.

minor comments (5)

[Abstract and Section III-D] The paper inconsistently refers to 'MobileNet' and 'MobileNetV2'; please settle on one name and use it consistently throughout.
[Section III-C] The architecture description states that MobileNet starts with 'fully convolutional layers with 32 filters and includes 19 residual bottleneck layers,' which describes MobileNetV2 rather than the original MobileNet; please verify the description against the cited reference.
[Section IV-A] The heading 'Experment Setup and Training Options' contains a typo; please change to 'Experiment Setup and Training Options.'
[Section II] Some related-work entries (e.g., references [17] and [18]) appear to be self-citations that are not clearly related to retinal disease classification; please either connect them to the narrative or remove them.
[Section III-A] The paper states that the data was 'randomly divided to ensure a balanced distribution' but Table I shows that the Normal class is larger than the DR and MH classes; please clarify whether any stratification or class balancing was performed.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: the reported accuracies are empirical test-set measurements, not predictions derived from fitted parameters or self-citation.

full rationale

The paper's central claim is that MobileNetV2 achieved 90.8% accuracy and NASNetMobile 89.5% on a three-class subset (Normal, DR, MH) of the external RFMiD dataset. These figures are measured outcomes obtained by training the models on the training partition and evaluating on a held-out test partition; they are not computed from the same fitted parameters that define the models, nor are they presented as predictions inferred from a fitted input. The architectures are off-the-shelf pretrained CNNs with the classification head replaced, and no mathematical derivation chain equates the reported accuracy to the experimental setup by construction. Comparisons in Table III are against independently published accuracy values on the same dataset, not quantities derived from this paper's own inputs. Some citations, including [3], [4], [17], and [18], involve overlapping authors, but they support general background about CNNs and deep learning rather than a load-bearing uniqueness claim or a forbidden alternative. The skeptic's concerns about the undocumented subset selection from the 46-class multi-label RFMiD, the single random split without patient-level stratification, and the small 362-image test set are validity and reproducibility issues, not circularity: an insufficiently described benchmark does not make the measured accuracy definitionally equal to a fitted parameter. Therefore, no circular step is exhibited, and the appropriate finding is no significant circularity.

Assumptions & free parameters 6 free parameters · 5 assumptions · 0 invented entities

The central claim depends on external dataset labels, the random split, pretrained ImageNet weights, and hand-set hyperparameters. The paper introduces no new entities, forces, or mechanisms. The main burden is empirical trust in the data and the split rather than theoretical derivation.

free parameters (6)

learning_rate = 0.001
Chosen in Section IV-A with no search or sensitivity analysis.
batch_size = 32
Set in Section IV-A without justification.
epochs = 30
Training length fixed at 30 with no early stopping or selection rationale.
dropout_rate = 0.3
Dropout in the custom head, chosen without analysis.
dense_units = 1024
Fully connected layer width chosen without optimization.
augmentation_parameters = rotation +/-30 degrees, shifts and shear up to 20%, zoom up to 20%, horizontal flip
Augmentation ranges are set by hand in Section III-B and not justified.

assumptions (5)

domain assumption RFMiD labels are correct and the selected three-class subset is a valid benchmark.
The paper trusts ground truth and selects Normal, DR, and MH from a 46-class multi-label dataset without auditing multi-disease images (Section III-A).
domain assumption The random train/test split has no patient-level leakage and is representative.
Section III-A says the data were randomly divided, but no subject-level grouping or repeated splits are described.
domain assumption ImageNet-pretrained weights transfer useful features to fundus images.
Transfer learning from ImageNet is assumed beneficial in Sections I and II without architecture-specific justification.
domain assumption Data augmentation transformations preserve the disease label.
Rotations, shifts, zooms, and flips are applied in Section III-B without verifying label invariance for these retinal conditions.
standard math Standard CNN and TensorFlow/Keras operations behave as documented.
Model training relies on standard libraries and no formal verification is provided.

how reviews work

0 comments

Cite this review

Pith. "Pith review of Lightweight Convolutional Neural Networks for Retinal Disease Classification." pith.science (2026). https://pith.science/paper/SS2RRSAB

@misc{pith2026250603186,
  author       = {Pith},
  title        = {Pith review of: Lightweight Convolutional Neural Networks for Retinal Disease Classification},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/SS2RRSAB}},
  note         = {Machine review of arXiv:2506.03186}
}

read the original abstract

Retinal diseases such as Diabetic Retinopathy (DR) and Macular Hole (MH) significantly impact vision and affect millions worldwide. Early detection is crucial, as DR, a complication of diabetes, damages retinal blood vessels, potentially leading to blindness, while MH disrupts central vision, affecting tasks like reading and facial recognition. This paper employed two lightweight and efficient Convolution Neural Network architectures, MobileNet and NASNetMobile, for the classification of Normal, DR, and MH retinal images. The models were trained on the RFMiD dataset, consisting of 3,200 fundus images, after undergoing preprocessing steps such as resizing, normalization, and augmentation. To address data scarcity, this study leveraged transfer learning and data augmentation techniques, enhancing model generalization and performance. The experimental results demonstrate that MobileNetV2 achieved the highest accuracy of 90.8%, outperforming NASNetMobile, which achieved 89.5% accuracy. These findings highlight the effectiveness of CNNs in retinal disease classification, providing a foundation for AI-assisted ophthalmic diagnosis and early intervention.

Figures

Figures reproduced from arXiv: 2506.03186 by the authors.

Discussion (0). Sign in to comment.

Reference graph

Works this paper leans on

24 extracted references · 20 canonical work pages

[1]

Development and validation of a deep learning algorithm for detection of diabetic retinopathy in retinal fundus photographs,

V. Gulshan et al., “Development and validation of a deep learning algorithm for detection of diabetic retinopathy in retinal fundus photographs,” jama, vol. 316, no. 22, pp. 2402–2410, 2016

work page 2016
[2]

Classification of multiple retinal disorders from enhanced fundus images using semi -supervised GAN,

A. Smitha and P. Jidesh, “Classification of multiple retinal disorders from enhanced fundus images using semi -supervised GAN,” SN Comput. Sci., vol. 3, no. 1, p. 59, 2022

work page 2022
[3]

Artificial intelligence based -improving reservoir management: An Attention -Guided Fusion Model for predicting injector –producer connectivity,

A. Saihood, T. Saihood, S. A. Jebur, C. Ehlig -Economides, L. Alzubaidi, and Y. Gu, “Artificial intelligence based -improving reservoir management: An Attention -Guided Fusion Model for predicting injector –producer connectivity,” Eng. Appl. Artif. Intell., vol. 146, p. 110205, 2025

work page 2025
[4]

Employing the Concept of Stacking Ensemble Learning to Generate Deep Dream Images Using Multiple CNN Variants,

L. Alkhazraji et al. , “Employing the Concept of Stacking Ensemble Learning to Generate Deep Dream Images Using Multiple CNN Variants,” Intell. Syst. with Appl., p. 200488, 2025

work page 2025
[5]

A Comparative Study of IDS-Based Deep Learning Models for IoT Network,

B. N. Shaker, B. Q. Al -Musawi, and M. F. Hassan, “A Comparative Study of IDS-Based Deep Learning Models for IoT Network,” in Proceedings of the 2023 International Conference on Advances in Artificial Intelligence and Applications, 2023, pp. 15–

work page 2023
[6]

Automatic classification of retinal diseases with transfer learning -based lightweight convolutional neural network,

Z. Lu, J. Miao, J. Dong, S. Zhu, X. Wang, and J. Feng, “Automatic classification of retinal diseases with transfer learning -based lightweight convolutional neural network,” Biomed. Signal Process. Control, vol. 81, p. 104365, 2023

work page 2023
[7]

Osteoporosis detection using convolutional neural network based on dual -energy X -ray absorptiometry images,

A. Z. Mohammed and L. E. George, “Osteoporosis detection using convolutional neural network based on dual -energy X -ray absorptiometry images,” Indones. J. Electr. Eng. Comput. Sci., vol. 29, no. 1, pp. 315 –321, 2022, doi: 10.11591/ijeecs.v29.i1.pp315- 321

work page doi:10.11591/ijeecs.v29.i1.pp315- 2022
[8]

Optimized deep convolutional neural networks for identification of macular diseases from optical coherence tomography images,

Q. Ji, J. Huang, W. He, and Y. Sun, “Optimized deep convolutional neural networks for identification of macular diseases from optical coherence tomography images,” Algorithms, vol. 12, no. 3, pp. 1– 12, 2019, doi: 10.3390/a12030051

work page doi:10.3390/a12030051 2019

Show all 24 references

[9]

A convolutional neural network for the screening and staging of diabetic retinopathy,

M. Shaban et al. , “A convolutional neural network for the screening and staging of diabetic retinopathy,” PLoS One, vol. 15, no. 6, p. e0233514, 2020. Mobile NetV2 NASNet Mobile

2020
[10]

Classification of retinal images based on convolutional neural network,

N. A. El‐Hag et al. , “Classification of retinal images based on convolutional neural network,” Microsc. Res. Tech., vol. 84, no. 3, pp. 394–414, 2021

2021
[11]

Retina diseases diagnosis using deep learning,

A. A. E. F. Elsharif and S. S. Abu -Naser, “Retina diseases diagnosis using deep learning,” 2022

2022
[12]

Multi categorical of common eye disease detect using convolutional neural network: a transfer learning approach,

A. K. Bitto and I. Mahmud, “Multi categorical of common eye disease detect using convolutional neural network: a transfer learning approach,” Bull. Electr. Eng. Informatics, vol. 11, no. 4, pp. 2378–2387, 2022

2022
[13]

Explainable Convolutional Neural Networks for Retinal Fundus Classification and Cutting-Edge Segmentation Models for Retinal Blood Vessels from Fundus Images,

F. T. J. Faria, M. Bin Moin, P. Debnath, A. I. Fahim, and F. M. Shah, “Explainable Convolutional Neural Networks for Retinal Fundus Classification and Cutting-Edge Segmentation Models for Retinal Blood Vessels from Fundus Images,” arXiv Prepr. arXiv2405.07338, 2024

2024 arXiv
[14]

Deep Learning Based Multi -Class Eye Disease Classification: Enhancing Vision Health Diagnosis,

J. Aslam, M. A. Arshed, S. Iqbal, and H. M. Hasnain, “Deep Learning Based Multi -Class Eye Disease Classification: Enhancing Vision Health Diagnosis,” Tech. J., vol. 29, no. 01, pp. 7–12, 2024

2024
[15]

Instantaneous Classification and Localization of Eye Diseases via Artificial Intelligence,

J. Biswas, S. M. Mustaquim, S. M. S. Hossain, and I. M. Siddique, “Instantaneous Classification and Localization of Eye Diseases via Artificial Intelligence,” Eur. J. Adv. Eng. Technol., vol. 11, no. 3, pp. 45–53, 2024

2024
[16]

Retinal fundus multi -disease image dataset (rfmid): A dataset for multi-disease detection research,

S. Pachade et al. , “Retinal fundus multi -disease image dataset (rfmid): A dataset for multi-disease detection research,” Data, vol. 6, no. 2, p. 14, 2021

2021
[17]

A Scalable and Generalised Deep Learning Framework for Anomaly Detection in Surveillance Videos,

S. A. Jebur, L. Alzubaidi, A. Saihood, K. A. Hussein, H. K. Hoomod, and Y. Gu, “A Scalable and Generalised Deep Learning Framework for Anomaly Detection in Surveillance Videos,” Int. J. Intell. Syst., vol. 2025, no. 1, p. 1947582, 2025

2025
[18]

ATD Learning: A secure, smart, and decentralised learning method for big data environments,

L. Alzubaidi et al. , “ATD Learning: A secure, smart, and decentralised learning method for big data environments,” Inf. Fusion, p. 102953, 2025

2025
[19]

Heart Attack Prediction by Integrating Independent Component Analysis with Machine Learning Classifiers,

A. K. Nawar et al. , “Heart Attack Prediction by Integrating Independent Component Analysis with Machine Learning Classifiers,” Res. Sq., 2024, doi: https://doi.org/10.21203/rs.3.rs - 5256555/v1

2024 doi
[20]

Multi -label classification of fundus images with efficientnet,

J. Wang, L. Yang, Z. Huo, W. He, and J. Luo, “Multi -label classification of fundus images with efficientnet,” IEEE access, vol. 8, pp. 212499–212508, 2020

2020
[21]

doi: 10.1145/3603273.3635058
[22]

ViLReF: An Expert Knowledge Enabled Vision - Language Retinal Foundation Model,

S. Yang et al., “ViLReF: An Expert Knowledge Enabled Vision - Language Retinal Foundation Model,” arXiv Prepr. arXiv2408.10894, 2024

2024
[23]

A deep learning framework for the early detection of multi-retinal diseases,

S. Ejaz, R. Baig, Z. Ashraf, M. M. Alnfiai, M. M. Alnahari, and R. M. Alotaibi, “A deep learning framework for the early detection of multi-retinal diseases,” PLoS One , vol. 19, no. 7, p. e0307317, 2024

2024
[24]

Deep hybrid architecture with stacked ensemble learning for binary classification of retinal disease,

C. Priyadharsini, “Deep hybrid architecture with stacked ensemble learning for binary classification of retinal disease,” Results Eng., vol. 24, p. 103219, 2024

2024

Pith tools

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

[1] [1]

Development and validation of a deep learning algorithm for detection of diabetic retinopathy in retinal fundus photographs,

V. Gulshan et al., “Development and validation of a deep learning algorithm for detection of diabetic retinopathy in retinal fundus photographs,” jama, vol. 316, no. 22, pp. 2402–2410, 2016

work page 2016

[2] [2]

Classification of multiple retinal disorders from enhanced fundus images using semi -supervised GAN,

A. Smitha and P. Jidesh, “Classification of multiple retinal disorders from enhanced fundus images using semi -supervised GAN,” SN Comput. Sci., vol. 3, no. 1, p. 59, 2022

work page 2022

[3] [3]

Artificial intelligence based -improving reservoir management: An Attention -Guided Fusion Model for predicting injector –producer connectivity,

A. Saihood, T. Saihood, S. A. Jebur, C. Ehlig -Economides, L. Alzubaidi, and Y. Gu, “Artificial intelligence based -improving reservoir management: An Attention -Guided Fusion Model for predicting injector –producer connectivity,” Eng. Appl. Artif. Intell., vol. 146, p. 110205, 2025

work page 2025

[4] [4]

Employing the Concept of Stacking Ensemble Learning to Generate Deep Dream Images Using Multiple CNN Variants,

L. Alkhazraji et al. , “Employing the Concept of Stacking Ensemble Learning to Generate Deep Dream Images Using Multiple CNN Variants,” Intell. Syst. with Appl., p. 200488, 2025

work page 2025

[5] [5]

A Comparative Study of IDS-Based Deep Learning Models for IoT Network,

B. N. Shaker, B. Q. Al -Musawi, and M. F. Hassan, “A Comparative Study of IDS-Based Deep Learning Models for IoT Network,” in Proceedings of the 2023 International Conference on Advances in Artificial Intelligence and Applications, 2023, pp. 15–

work page 2023

[6] [6]

Automatic classification of retinal diseases with transfer learning -based lightweight convolutional neural network,

Z. Lu, J. Miao, J. Dong, S. Zhu, X. Wang, and J. Feng, “Automatic classification of retinal diseases with transfer learning -based lightweight convolutional neural network,” Biomed. Signal Process. Control, vol. 81, p. 104365, 2023

work page 2023

[7] [7]

Osteoporosis detection using convolutional neural network based on dual -energy X -ray absorptiometry images,

A. Z. Mohammed and L. E. George, “Osteoporosis detection using convolutional neural network based on dual -energy X -ray absorptiometry images,” Indones. J. Electr. Eng. Comput. Sci., vol. 29, no. 1, pp. 315 –321, 2022, doi: 10.11591/ijeecs.v29.i1.pp315- 321

work page doi:10.11591/ijeecs.v29.i1.pp315- 2022

[8] [8]

Optimized deep convolutional neural networks for identification of macular diseases from optical coherence tomography images,

Q. Ji, J. Huang, W. He, and Y. Sun, “Optimized deep convolutional neural networks for identification of macular diseases from optical coherence tomography images,” Algorithms, vol. 12, no. 3, pp. 1– 12, 2019, doi: 10.3390/a12030051

work page doi:10.3390/a12030051 2019