With the explosive growth of network data, data privacy protection has become a focus for many researchers. To better balance privacy protection and data utility and strengthen defense during model training, a Federated Learning (FL) method based on Differential Privacy (DP) is proposed. This method evaluates the weights uploaded by participants by constructing a scoring function. The random sampling based on an exponential mechanism is introduced to enhance dat a privacy. The proposed framework leverages federated learning and differential privacy to mitigate the adverse impact of attackers and untrusted par ties. From the experimental results, the image quality reached the highest on the Structural Similarity Ind ex (SSIM) , at 0.7, while the multi -scale structural similarity was 0.65, demonstrating strong visualization quality. When the participant was 100, the classification accuracy reached 83.5%. In addition, the classification accuracy was 90.2% when the privac y budget was 3. The research has proven that the proposed method can effectively improve image quality and classification accuracy while protecti ng image data privacy, providing strong support for enterprises and units to legally and efficiently utilize da ta.

[1] Awan K., Din I., Almogren A., and Rodrigues J., “Privacy -Preservi ng Big Data Security for Iot with Federated Learning and Cryptography, ” IEEE Access , vol. 11, pp. 120918 -120934. 2023. DOI: 10.1109/ACCESS.2023.3328310

[2] Bandara F., Jayawickrama U., Subasinghage M., Olan F, and et al., “Enhancing ERP Responsiveness Through Big Data Technologies: An Empirical Investigation ” Information Systems Frontiers , vol. 26, no. 1, pp. 251 -275, 2024. https://doi.org/10.1007/s10796 -023 -10374 -w

[3] Beikmohammadi A., Khirirat S., and Magnusson S., “On the Convergence of Federated Learning Algor ithms Without Data Similarity, ” IEEE Transactions on Big Data , vol. 11, no. 2, pp. 659 - 668, 2024. DOI: 10.1109/TBDATA.2024.3423693

[4] Cao Y., Wei W., and Zhou J., “Privacy Protection Data Mining Algorithm in Blockchain Based on Decision Tree Classification, ” Web Intelligence , vol. 20, no. 2, pp. 103 -112, 2022. https://doi.org/10.3233/WEB -210485

[5] Demirol D., Das R., and Hanbay D., “A Key Review on Security and Privacy of Big Data: Issues, Challenges, and Future Research Directions, ” Signal, Image and Video Proce ssing , vol. 17, no. 4, pp. 1335 -1343, 2023. https://doi.org/10.1007/s11760 -022 -02341 -w

[6] Dong J., Roth A., and Su W J., “Gaussian differential privacy, ” Journal of the Royal Statistical Society Series B: Statistical Methodology , vol. 84, no. 1, pp. 3 -37, 202 2. https://doi.org/10.1111/rssb.12454

[7] Goncalves C., Bessa R., and Pinson P., “Privacy - Preserving Distributed Learning for Renewable Energy Forecasting, ” IEEE Transactions on Sustainable Energy , vol. 12, no. 3, pp. 1777 -1787, 2021. DOI: 10.1109/TSTE.2021.30 65117

[8] He Y., Ouyang W., Li S., Wang L., and et al., “Cloud Computing Data Privacy Protection Method Based on Blockchain, ” International Journal of Grid and Utility Computing , vol. 14, no. 5, pp. 480 -492, 2023. https://doi.org/10.1504/ijguc.2023.133436

[9] Hida yat M., Nakamura Y., and Arakawa Y., “Privacy -Preserving Federated Learning with Resource -Adaptive Compression for Edge Devices, ” IEEE Internet of Things Journal , vol. 11, no. 8, pp. 13180 -13198, 2024. DOI: 10.1109/JIOT.2023.3347552

[10] Kerkouche R., Acs G., C astelluccia C., Geneves P., “Compression Boosts Differentially Private Federated Learning, ” in Proceedings of the IEEE European Symposium on Security and Privacy , Vienna, pp. 304 -318, 2021. DOI: 10.1109/EuroSP51992.2021.00029

[11] Koley S., Roy H., Dhar S., and Bhattacharjee D., “Cross -Modal Face Recognition with Illumination -Invariant Local Discrete Cosine Transform Binary Pattern (LDCTBP), ” Pattern Analysis and Applications , vol. 26, no. 3, pp. 847 - 859, 2023. https://doi.org/10.1007/s10044 -023 - 01139 -x

[12] Liu R., Cao Y., Yoshikawa M., and Chen H., “FedSel: Federated SGD under Local Differential Privacy with Top -k Dimension Selection, ” in Proceedings of the 25 th International Conference on Database Systems for Advanced Applications , Jeju, pp. 485 -501, 2020. https:// link.springer.com/chapter/10.1007/978 -3- 030 -59410 -7_33

[13] McMahan H., Ramage D., Talwar K., and Zhang L., “Learning Differentially Private Recurrent Language Models, ” arXiv preprint , arXiv:1710.06963v3, pp. 1 -14, 2017. https://doi.org/10.48550/arXiv.1710.0696 3

[14] Miao Y., Xie R., Li X., Liu X., and et al, “Compressed Federated Learning Based on Adaptive Local Differential Privacy, ” in Proceedings of the 38 th Annual Computer Security Applications Conference , Texas, pp. 159 -170, 2022. https://doi.org/10.1145/356462 5.3567973

[15] Pei J., Liu W., Li J., Wang L., and Liu C., “A Review of Federated Learning Methods in Heterogeneous Scenarios, ” IEEE Transactions on Consumer Electronics , vol. 70, no. 3, pp. 5983 - 5999. 2024. DOI: 10.1109/TCE.2024.3385440

[16] Shao Z., Wang X., Tang Y., and Shang Y., “Trinion Discrete Cosine Transform with Application to Color Image Encryption, ” Multimedia Tools and Applications , vol. 82, no. 10, pp. 14633 -14659, 2023. https://doi.org/10.1007/s11042 -022 -13898 -6

[17] Thakkar O., Ramaswamy S., Mathews R., an d Beaufays F., “Understanding Unintended Memorization in Federated Learning, ” arXiv preprint , arXiv:2006.07490v1, pp. 1 -14, 2020. https://doi.org/10.48550/arXiv.2006.07490

[18] Tunc -Abubakar T., Kalkan A., and Abubakar A., “Impact of Big Data Usage on Product a nd Process Innovation: The Role of Data Diagnosticity, ” Kybernetes , vol. 52, no. 9, pp. 3178 -3196, 2023. https://doi.org/10.1108/K -11 - 2021 -1138

[19] Yanamala A., “Secure and Private AI: Implementing Advanced Data Protection Techniques in Machine Learning Models ,” International Journal of Machine Learning Research in Cybersecurity and Artificial Intelligence , vol. 14, no. 1, pp. 105 -132, 2023. https://scholar.google.com/citations?view_op=vie w_citation&hl=en&user=aivJDSwAAAAJ&citat ion_for_view=aivJDSwAAAAJ:qjMakFH Dy7sC

[20] Zhao L., Wang Q., Zou Q., Zhang Y., and Chen Y., “Privacy -Preserving Collaborative Deep Learning with Unreliable Participants, ” IEEE Transactions on Information Forensics and Security , vol. 15, pp. 1486 -1500, 2020. DOI: 10.1109/TIFS.2019.2939713