pith. sign in

arxiv: 2605.27816 · v1 · pith:Y2D3TLBGnew · submitted 2026-05-27 · 💻 cs.CV

Pattern Recognition Tasks with Personalized Federated Learning

Md. Arifur Rahman , Isha Das , Mushfiqur Rahman Abir , B. M. Taslimul Haque , Abdullah Al Noman , Abir Ahmed , Md. Jakir Hossen This is my paper

Pith reviewed 2026-06-29 14:04 UTC · model grok-4.3

classification 💻 cs.CV
keywords personalized federated learningpattern recognitioncomparative analysisheterogeneous dataMNISTSignMNISTDigit5performance metrics
0
0 comments X

The pith

APPLE, FedGC, and FedProto outperform other personalized federated learning methods on pattern recognition tasks with heterogeneous client data.

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

The paper compares seven personalized federated learning algorithms across three image datasets for pattern recognition tasks. It evaluates them using accuracy, precision, recall, and F1 score to find which ones best adapt models to individual clients while keeping data private. The results identify APPLE, FedGC, and FedProto as the strongest performers that maintain high accuracy across MNIST, SignMNIST, and Digit5. The study also outlines how each algorithm operates along with their specific advantages and limitations in handling varied data distributions.

Core claim

Through empirical investigation on MNIST, SignMNIST, and Digit5, the evaluation anchored in Accuracy, Precision, Recall, and F1 Score shows that APPLE, FedGC, and FedProto consistently deliver superior performance in personalized federated learning for pattern recognition tasks, with other algorithms showing advantages only in specific contexts and with potential for further refinement.

What carries the argument

Comparative evaluation of seven PFL algorithms on heterogeneous client data distributions using four standard classification metrics.

If this is right

  • APPLE, FedGC, and FedProto can be prioritized for pattern recognition applications that involve client-specific data distributions.
  • Algorithm selection in PFL requires attention to the particular dataset and data heterogeneity patterns.
  • Context-specific alternatives among the seven algorithms remain viable when datasets differ from those tested.
  • Iterative refinement of the top methods can further improve outcomes in privacy-preserving settings.

Where Pith is reading between the lines

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

  • The superiority of these three algorithms may extend to other image-based tasks with similar privacy needs, though this requires direct testing.
  • Limiting the study to three datasets leaves open whether the ranking holds for non-image pattern recognition problems.
  • The results point toward using these methods as baselines when designing new PFL systems for heterogeneous client environments.

Load-bearing premise

The three chosen datasets and four metrics are representative and sufficient to identify the best PFL algorithms for pattern recognition tasks with heterogeneous client data.

What would settle it

An experiment on a new heterogeneous image dataset for pattern recognition where one of the other four PFL algorithms records higher average scores across accuracy, precision, recall, and F1 than APPLE, FedGC, and FedProto.

Figures

Figures reproduced from arXiv: 2605.27816 by Abdullah Al Noman, Abir Ahmed, B. M. Taslimul Haque, Isha Das, Md. Arifur Rahman, Md. Jakir Hossen, Mushfiqur Rahman Abir.

Figure 1
Figure 1. Figure 1: Framework Workflow 3-1-Overview of Datasets 3-1-1- MNIST The MNIST dataset is a widely used benchmark dataset in the field of ML and pattern recognition. It comprises a collection of 70,000 grayscale images of handwritten digits (0 to 9), each having a resolution of 28x28 pixels. The dataset is divided into two main subsets: 60,000 images for training and 10,000 images for testing. Its simplicity and avail… view at source ↗
read the original abstract

Personalized Federated Learning (PFL) constitutes a novel paradigm that tailors Machine Learning (ML) models to individual clients, thereby furnishing personalized model updates whilst upholding stringent data privacy principles. Diverging from conventional standard Federated Learning (FL) approaches, PFL adapts models to distinct client data distributions, engendering heightened levels of accuracy, customization, and data security, all while minimizing communication overhead. This methodology proves particularly salient in contexts marked by pattern recognition tasks reliant upon heterogeneous data sources and underpinned by paramount privacy apprehensions. In the present research endeavor, this article undertake a comprehensive comparative analysis of seven distinct PFL algorithms deployed across three diverse datasets, namely MNIST, SignMNIST, and Digit5. The overarching objective entails ascertaining the preeminent PFL algorithm, within the framework of pattern recognition tasks, through a rigorous evaluation anchored in metrics encompassing Accuracy, Precision, Recall, and F1 Score. Concurrently, an in-depth scrutiny of these PFL algorithms is conducted, elucidating their operative workflows, advantages, and limitations. Through empirical investigation, the findings evince that APPLE, FedGC, and FedProto emerge as stalwart contenders, consistently furnishing superior performance across the spectrum of assessed datasets, while acknowledging the contextual specificity of alternative algorithms and the potential for iterative refinement to realize optimal outcomes.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Referee Report

2 major / 2 minor

Summary. The paper performs an empirical comparison of seven Personalized Federated Learning (PFL) algorithms on three image classification datasets (MNIST, SignMNIST, Digit5). It evaluates them using Accuracy, Precision, Recall, and F1 Score, analyzes their workflows/advantages/limitations, and concludes that APPLE, FedGC, and FedProto consistently deliver superior performance across the assessed datasets for pattern recognition tasks under heterogeneous client data.

Significance. If the empirical comparisons are made reproducible with full protocol details and statistical rigor, the work supplies a useful reference benchmark for PFL algorithm selection in privacy-sensitive image classification with non-IID data. The side-by-side discussion of each method's mechanics adds practical value. The narrow dataset scope, however, limits the strength of any claim to identify preeminent methods for the broader class of pattern recognition tasks.

major comments (2)
  1. [Experimental Setup / Results] Experimental Setup / Results sections: The manuscript asserts empirical superiority of APPLE, FedGC, and FedProto but supplies no details on data partitioning for heterogeneity, number of clients, communication rounds, hyperparameter selection, baseline implementations, or statistical tests. This information is required to verify the central performance claims.
  2. [Abstract and Conclusion] Abstract and Conclusion: The claim that the three algorithms are preeminent 'within the framework of pattern recognition tasks' rests on results from only MNIST, SignMNIST, and Digit5 (all digit/sign variants with similar input statistics). No justification or additional domains are provided to support extrapolation beyond these specific datasets.
minor comments (2)
  1. [Abstract] Abstract contains grammatical errors (e.g., 'this article undertake' should be 'undertakes'; 'engendering heightened levels of accuracy' is awkward).
  2. No mention of code or data availability, which would strengthen reproducibility of the reported comparisons.

Simulated Author's Rebuttal

2 responses · 0 unresolved

We thank the referee for the constructive comments. We address each major comment point by point below and will revise the manuscript to improve reproducibility and appropriately scope the claims.

read point-by-point responses

  1. Referee: Experimental Setup / Results sections: The manuscript asserts empirical superiority of APPLE, FedGC, and FedProto but supplies no details on data partitioning for heterogeneity, number of clients, communication rounds, hyperparameter selection, baseline implementations, or statistical tests. This information is required to verify the central performance claims.

    Authors: We agree that the current version lacks these protocol details, which are necessary for reproducibility. In the revised manuscript we will add a dedicated subsection in Experimental Setup describing: the non-IID partitioning method (Dirichlet distribution with chosen alpha), number of clients and per-client sample counts, total communication rounds, hyperparameter selection procedure (including values used or tuning method), baseline implementation sources or re-implementation details, and statistical significance tests (e.g., paired t-tests or Wilcoxon signed-rank with p-values) supporting the reported superiority. These additions will be included in the next version. revision: yes

  2. Referee: Abstract and Conclusion: The claim that the three algorithms are preeminent 'within the framework of pattern recognition tasks' rests on results from only MNIST, SignMNIST, and Digit5 (all digit/sign variants with similar input statistics). No justification or additional domains are provided to support extrapolation beyond these specific datasets.

    Authors: We accept that the three datasets share similar input statistics and that broad extrapolation is not justified. We will revise the abstract and conclusion to state that APPLE, FedGC, and FedProto demonstrate superior performance on the evaluated image-classification datasets for pattern recognition tasks, without claiming preeminence across all pattern recognition tasks. A new limitations paragraph will explicitly note the narrow dataset scope and recommend future validation on more diverse domains. revision: yes

Circularity Check

0 steps flagged

No circularity: purely empirical comparison of PFL algorithms

full rationale

The paper performs an empirical comparison of seven PFL algorithms on MNIST, SignMNIST, and Digit5 using Accuracy, Precision, Recall, and F1 Score. It reports observed performance rankings without any derivation chain, first-principles predictions, fitted parameters renamed as predictions, or load-bearing self-citations. The central claim is that APPLE, FedGC, and FedProto perform best on the tested datasets; this is a direct reporting of experimental outcomes rather than a reduction to inputs by construction. No equations or uniqueness theorems are invoked that could create circularity. The analysis is self-contained against external benchmarks (the three datasets and four metrics).

Axiom & Free-Parameter Ledger

0 free parameters · 0 axioms · 0 invented entities

No mathematical derivations, free parameters, axioms, or invented entities; the work is an empirical benchmark study.

pith-pipeline@v0.9.1-grok · 5790 in / 1018 out tokens · 39628 ms · 2026-06-29T14:04:16.563703+00:00 · methodology

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

54 extracted references · 32 canonical work pages · 1 internal anchor

  1. [1]

    DP-SVM MNIST Linear SVM = 97.2, RBF SVM = 98.3 The core contribution of the scientific paper is the development of a privacy -preserving image pattern recognition scheme using Support Vector Machine (SVM) and Differential Privacy (DP) framework

  2. [2]

    CNN Sign- MNIST Accuracy = 93 The core contribution of the scientific paper is the development of a DL based application that uses a custom Convolutional Neural Network (CNN) to recognize sign language gestures from video frames

  3. [3]

    The paper compares the performance of ML and DL algorithms, specifically the CNN classifier, for accurately extracting and classifying handwritten letters and numbers from images

    KNN MNIST Sensitivity = 97.18, Precision = 97.92, Specificity = 98.58 The core contribution of the scientific paper is the development and evaluation of a system for Optical Character Recognition (OCR) focused on handwritten digits. The paper compares the performance of ML and DL algorithms, specifically the CNN classifier, for accurately extracting and c...

  4. [4]

    Decision Tree MNIST True-Classified Scenario = 90.37, False-Classified Scenario = 9.63 The core contribution of the scientific paper is the proposal of a method for offline handwritten digit recognition using ML algorithms. Various ML methods, such as Support Vector Machine, Multilayer Perceptron, Decision Tree, Naïve Bayes, K-Nearest Neighbor, and Random...

  5. [5]

    GPFL uses multiple GP runs to generate models that focus on specific high -level features of tr aining images

    GPFL MNIST Accuracy = 91.66 The core contribution of the scientific paper is the introduction of Genetic Program Feature Learner (GPFL), a novel generative GP feature learner for 2D images. GPFL uses multiple GP runs to generate models that focus on specific high -level features of tr aining images. These models are then combined into a function that reco...

  6. [6]

    Instead of using the gradient algorithm to adjust CNN parameters, the model inputs CNN-extracted features into RF for classification

    Hybrid-RF MNIST Error = 1.97 The core contribution of the scientific paper is the proposal of a hybrid model that combines a Convolutional Neural Network (CNN) and a Random Forest (RF) for image pattern recognition. Instead of using the gradient algorithm to adjust CNN parameters, the model inputs CNN-extracted features into RF for classification. Emergin...

  7. [7]

    FedAvg MNIST, Shakespeare, CIFAR-10 Accuracy: MNIST=99, Shakespeare=54, CIFAR- 10=85 The core contribution of this paper is the introduction of FL, a decentralized method for training deep networks on distributed mobile data, minimizing communication costs while ensuring strong performance across diverse data distributions

  8. [8]

    SCAFFOLDE MNIST Similarity=84.2 The core contribution of this paper is the SCAFFOLD algorithm, which addresses the ’client -drift’ issue in FL by using control variates for variance reduction, resulting in faster and more stable convergence, particularly in the presence of data heterogeneity and client sampling

  9. [9]

    FedProx MNIST, FEMNIST, Shakespeare, Sent140 Average accuracy is more than 22% than FedAvg The core contribution of this paper is the introduction of FedProx, a novel FL framework that handles statistical and systemic heterogeneity, ensures convergence for non- identical data distributions, and outperforms state-of-the- art FedAvg in robustness and accuracy

  10. [10]

    FedDyn MNIST, EMNIST, CIFAR-10, CIFAR-100 and Shakespeare Accuracy: MNIST=98.4, EMNIST=95.0, CIFAR-10=85.2, CIFAR-100=55.3, Shakespeare=51.2 The core contribution of this study is a new FL approach that dynamically regulates the interoperation between a subset of devices, coordinating local and global solutions while minimizing transmission costs and taki...

  11. [11]

    MOON CIFAR-10, CIFAR-100, and Tiny- Imagenet True-Classified Scenario = 90.37, False-Classified Scenario = 9.63 The core contribution of this paper is the introduction of MOON, a model - contrastive FL approach that leverages model representation similarity to enhance federated training, achieving superior performance on image pattern recognition tasks co...

  12. [12]

    3- Research Methodology FL algorithms were implemented by first defining the problem intended to be solved through this distributed approach

    FedGen MNIST, EMNIST, Celeba Accuracy: MNIST=95.52, EMNIST=78.92, Celeba=90.29 The core contribution of this paper is a data-free knowledge distillation approach that enhances heterogeneous FL by utilizing a lightweight generator to ensemble user information, leading to improved generalization performance and reduced communication rounds compared to state...

    2023

  13. [13]

    S., & Reed, J

    Niknam, S., Dhillon, H. S., & Reed, J. H. (2020). Federated Learning for Wireless Communications: Motivation, Opportunities, and Challenges. IEEE Communications Magazine, 58(6), 46–51. doi:10.1109/MCOM.001.1900461

    work page doi:10.1109/mcom.001.1900461 2020

  14. [14]

    T., & Morol, M

    Hosain, M. T., & Morol, M. K. (2025). Can Multi -turn Self -refined Single Agent LMs with Retrieval Solve Hard Coding Problems? Proceedings of the Annual Meeting of the Association for Computational Linguistics, 4, 129 –142. doi:10.18653/v1/2025.acl-srw.8

    work page doi:10.18653/v1/2025.acl-srw.8 2025

  15. [15]

    Voigt, P., & Von dem Bussche, A. (2017). The EU General Data Protection Regulation (GDPR): A Practical Guide. In The EU General Data Protection Regulation (GDPR): a Practical Guide. Springer International Publishing , Berlin, Germany . doi:10.1007/978-3-319-57959-7

    work page doi:10.1007/978-3-319-57959-7 2017

  16. [16]

    R., Mridha, M

    Abir, M. R., Mridha, M. F., & Shin, J. Federated Smart Mirrored Personalized Fashion in IoT . ResearchGate, 1 -5. doi:10.13140/RG.2.2.35779.52008

    work page doi:10.13140/rg.2.2.35779.52008

  17. [17]

    Brendan McMahan, H., Moore, E., Ramage, D., Hampson, S., & Agüera y Arcas, B. (2017). Communication -efficient learning of deep networks from decentralized data. Proceedings of the 20th International Conference on Artificial Intelligence and Statistics, AISTATS 2017, 54

    2017

  18. [18]

    T., Morol, M

    Hosain, M. T., Morol, M. K., & Hossen, M. J. (2025). A hybrid self-attentive linearized phrase structured transformer based RNN for financial sentence analysis with sentence level explainability. Scientific Reports, 15(1), 23893. doi:10.1038/s41598-025-09265-8

    work page doi:10.1038/s41598-025-09265-8 2025

  19. [19]

    Kairouz, P., & McMahan, H. B. (2021). Advances and open problems in federated learning. Foundations and trends in machine learning, 14(1-2), 1-210. doi:10.1561/2200000083

    work page doi:10.1561/2200000083 2021

  20. [20]

    Federated Learning with Personalization Layers

    Arivazhagan, M. G., Aggarwal, V., Singh, A. K., & Choudhary, S. (2019). Federated learning with personalization layers. arXiv Preprint, arXiv:1912.00818. doi:10.48550/arXiv.1912.00818

    work page internal anchor Pith review Pith/arXiv arXiv doi:10.48550/arxiv.1912.00818 2019

  21. [21]

    Mushfiqur Rahman Abir, Asif Zaman, & Sawon Mursalin. (2024). Efficiency measurement of FL algorithms for image classification. GSC Advanced Research and Reviews, 18(3), 356–366. doi:10.30574/gscarr.2024.18.3.0110

    work page doi:10.30574/gscarr.2024.18.3.0110 2024

  22. [22]

    P., Kale, S., Mohri, M., Reddi, S

    Karimireddy, S. P., Kale, S., Mohri, M., Reddi, S. J., Stich, S. U., & Suresh, A. T. (2020). SCAFFOLD: Stochastic Controlled Averaging for Federated Learning. 37th International Conference on Machine Learning, ICML 2020, PartF168147 -7, 5088– 5099

    2020

  23. [23]

    K., Zaheer, M., Sanjabi, M., Talwalkar, A., & Smith, V

    Li, T., Sahu, A. K., Zaheer, M., Sanjabi, M., Talwalkar, A., & Smith, V. (2020). Federated Optimization in Heterogeneous Networks. Proceedings of Machine Learning and Systems, 2, 429–450. Emerging Science Journal | Vol. 10, No. 2 Page | 989

    2020

  24. [24]

    Das, Hosain, M

    Gupta, R. Das, Hosain, M. T., Mridha, M. F., & Ahmed, S. U. (2025). Multimodal Programming in Computer Science with Interactive Assistance Powered by Large Language Model. Lecture Notes in Computer Science, 15770 LNCS, 59 –69. doi:10.1007/978-3-031-93864-1_5

    work page doi:10.1007/978-3-031-93864-1_5 2025

  25. [25]

    T., Rahman, S., Morol, M

    Hosain, M. T., Rahman, S., Morol, M. K., & Parvez, M. R. (2025). Xolver: Multi -agent reasoning with holistic experience learning just like an olympiad team. arXiv Preprint, arXiv:2506.14234. doi:10.48550/arXiv.2506.14234

    work page doi:10.48550/arxiv.2506.14234 2025

  26. [26]

    Theodoridis, S., & Koutroumbas, K. (2006). Pattern recognition. Elsevier, Amsterdam, Netherlands

    2006

  27. [27]

    Saini, P., Kaur, J., & Lamba, S. (2021). A Review on Pattern Recognition Using Machine Learning. Lecture Notes in Mechanical Engineering, 619–627. doi:10.1007/978-981-16-0942-8_58

    work page doi:10.1007/978-981-16-0942-8_58 2021

  28. [28]

    T., Gupta, R

    Hosain, M. T., Gupta, R. D., & Morol, M. K. (2025). Multilingual question answering in low -resource settings: A dzongkha - english benchmark for foundation models. arXiv Preprint, arXiv:2505.18638. doi:10.48550/arXiv.2505.18638

    work page doi:10.48550/arxiv.2505.18638 2025

  29. [29]

    Woubie, A., Solomon, E., & Attieh, J. (2024). Maintaining privacy in face recognition using federated learning method. IEEE Access, 12, 39603-39613. doi:10.1109/ACCESS.2024.3373691

    work page doi:10.1109/access.2024.3373691 2024

  30. [30]

    T., Abir, M

    Hosain, M. T., Abir, M. R., Rahat, M. Y., Mridha, M. F., & Mukta, S. H. (2024). Privacy Preserving Machine Learning with Federated Personalized Learning in Artificially Generated Environment. IEEE Open Journal of the Computer Society, 5, 694 –

    2024

  31. [31]

    doi:10.1109/OJCS.2024.3466859

    work page doi:10.1109/ojcs.2024.3466859 2024

  32. [32]

    T., Zaman, A., Sajid, M

    Hosain, M. T., Zaman, A., Sajid, M. S., Khan, S. S., & Akter, S. (2023). Privacy Preserving Machine Learning Model Personalization through Federated Personalized Learning. 2023 4th International Conference on Data Analytics for Business and Industry, ICDABI 2023, 536–545. doi:10.1109/ICDABI60145.2023.10629638

    work page doi:10.1109/icdabi60145.2023.10629638 2023

  33. [33]

    Luo, J., & Wu, S. (2022). Adapt to Adaptation: Learning Personalization for Cross-Silo Federated Learning. IJCAI International Joint Conference on Artificial Intelligence, 2166–2173. doi:10.24963/ijcai.2022/301

    work page doi:10.24963/ijcai.2022/301 2022

  34. [34]

    Zhang, J., Hua, Y., Wang, H., Song, T., Xue, Z., Ma, R., & Guan, H. (2023). FedALA: Adaptive Local Aggregation for Personalized Federated Learning. Proceedings of the 37th AAAI Conference on Artificial Intelligence, AAAI 2023, 37, 11237– 11244. doi:10.1609/aaai.v37i9.26330

    work page doi:10.1609/aaai.v37i9.26330 2023

  35. [35]

    Oh, J., Kim, S., & Yun, S. Y. (2021). Fedbabu: Towards enhanced representation for federated image classification. arXiv Preprint, arXiv:2106.06042. doi:10.48550/arXiv.2106.06042

    work page doi:10.48550/arxiv.2106.06042 2021

  36. [36]

    Niu, Y., & Deng, W. (2022). Federated Learning for Face Recognition with Gradient Correction. Proceedings of the 36th AAAI Conference on Artificial Intelligence, AAAI 2022, 36, 1999–2007. doi:10.1609/aaai.v36i2.20095

    work page doi:10.1609/aaai.v36i2.20095 2022

  37. [37]

    Xu, J., Tong, X., & Huang, S. L. (2023). Personalized federated learning with feature alignment and classifier collaboration. arXiv preprint arXiv:2306.11867. doi:10.48550/arXiv.2306.11867

    work page doi:10.48550/arxiv.2306.11867 2023

  38. [38]

    Tan, Y., Long, G., Ma, J., Liu, L., Zhou, T., & Jiang, J. (2022). Federated Learning from Pre -Trained Models: A Contrastive Learning Approach. Advances in Neural Information Processing Systems, 35, 19332–19344

    2022

  39. [39]

    Tan, Y., Long, G., Liu, L., Zhou, T., Lu, Q., Jiang, J., & Zhang, C. (2022). FedProto: Federated Prototype Learning across Heterogeneous Clients. Proceedings of the 36th AAAI Conference on Artificial Intelligence, AAAI 2022, 36, 8432 –8440. doi:10.1609/aaai.v36i8.20819

    work page doi:10.1609/aaai.v36i8.20819 2022

  40. [40]

    Deng, L. (2012). The MNIST database of handwritten digit images for machine learning research. IEEE Signal Processing Magazine, 29(6), 141–142. doi:10.1109/MSP.2012.2211477

    work page doi:10.1109/msp.2012.2211477 2012

  41. [41]

    El-Marzouki, N., Lasri, I., Riadsolh, A., & Elbelkacemi, M. (2024). American Sign Language Recognition with Convolutional Neural Networks: A Gateway to Enhanced Inclusivity. Lecture Notes in Networks and Systems, 1100 LNNS, 24 –33. doi:10.1007/978-3-031-68660-3_3

    work page doi:10.1007/978-3-031-68660-3_3 2024

  42. [42]

    Zhou, K., Yang, Y., Hospedales, T., & Xiang, T. (2020). Learning to Generate Novel Domains for Domain Generalization. Lecture Notes in Computer Science (Including Subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics), 12361 LNCS, 561–578. doi:10.1007/978-3-030-58517-4_33

    work page doi:10.1007/978-3-030-58517-4_33 2020

  43. [43]

    T., Jubair, M

    Abir, M., Hosain, M. T., Jubair, M. A. A., & Mridha, M. F. (2024). Imvb7t: A multi-modal model for food preferences based on artificially produced traits. Proceedings of the 3rd International Conference on Computing Advancements , 882-888. doi:10.1145/3723178.3723295

    work page doi:10.1145/3723178.3723295 2024

  44. [44]

    Senekane, M. (2019). Differentially Private Image Classification Using Support Vector Machine and Differential Privacy. Machine Learning and Knowledge Extraction, 1(1), 483–491. doi:10.3390/make1010029

    work page doi:10.3390/make1010029 2019

  45. [45]

    Nirmala, M. (2022). Sign Language Recognition Using Deep Learning. 2022 4th International Conference on Cognitive Computing and Information Processing, CCIP 2022, 12(5), 964–968. doi:10.1109/CCIP57447.2022.10058655. Emerging Science Journal | Vol. 10, No. 2 Page | 990

    work page doi:10.1109/ccip57447.2022.10058655 2022

  46. [46]

    Sharma, A., Bhardwaj, H., Bhardwaj, A., Sakalle, A., Acharya, D., & Ibrahim, W. (2022). A Machine Learning and Deep Learning Approach for Recognizing Handwritten Digits. Computational Intelligence and Neuroscience, 2022. doi:10.1155/2022/9869948

    work page doi:10.1155/2022/9869948 2022

  47. [47]

    K., Rajpurohit, J., & Abugabah, A

    Bhushan, S., Alshehri, M., Keshta, I., Chakraverti, A. K., Rajpurohit, J., & Abugabah, A. (2022). An Experimental Analysis of Various Machine Learning Algorithms for Hand Gesture Recognition. Electronics (Switzerland), 11(6) , 968 . doi:10.3390/electronics11060968

    work page doi:10.3390/electronics11060968 2022

  48. [48]

    Gope, B., Pande, S., Karale, N., Dharmale, S., & Umekar, P. (2021). Handwritten digits identification using mnist database via machine learning models. IOP Conference Series: Materials Science and Engineering , 1022(1), 012108. doi:10.1088/1757 - 899X/1022/1/012108

    work page doi:10.1088/1757 2021

  49. [49]

    Ruberto, S., Terragni, V., & Moore, J. H. (2020). Image feature learning with genetic programming. Lecture Notes in Computer Science (Including Subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics), 12270 LNCS, 63 –

    2020

  50. [50]

    doi:10.1007/978-3-030-58115-2_5

    work page doi:10.1007/978-3-030-58115-2_5

  51. [51]

    Xi, E. (2022). Image Classification and Recognition Based on Deep Learning and Random Forest Algorithm. Wireless Communications and Mobile Computing, 2022. doi:10.1155/2022/2013181

    work page doi:10.1155/2022/2013181 2022

  52. [52]

    E., Yue, Z., Ramon, M., Matthew, M., Paul, W., & Venkatesh, S

    Durmus, A. E., Yue, Z., Ramon, M., Matthew, M., Paul, W., & Venkatesh, S. (2021). Federated learning based on dynamic regularization. International Conference on Learning Representations, 10280833

    2021

  53. [53]

    Li, Q., He, B., & Song, D. (2021). Model-contrastive federated learning. Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, 10713–10722

    2021

  54. [54]

    Zhu, Z., Hong, J., & Zhou, J. (2021). Data-Free Knowledge Distillation for Heterogeneous Federated Learning. Proceedings of Machine Learning Research, 139, 12878–12889

    2021