Federated learning, as a distributed collaborative modeling framework, achieves joint learning under the premise of protecting data privacy by completing training locally and only sharing model updates. However, in practical applications, there are still risks such as gradient leakage and poisoning attacks. To enhance privacy protection and model robustness, this study proposes a protocol that combines differential privacy, identity-based signatures, ordered encryption and correlation detection to effectively identify and eliminate abnormal gradients. On this basis, an adaptive aggregation method based on local differential privacy is designed. Enable users to flexibly set privacy budgets and maintain the high precision of the global model by leveraging security aggregations. Theoretical analysis and experimental verification show that this method can not only effectively resist malicious attacks, but also improve the convergence and prediction performance of the model while ensuring privacy. It further demonstrates its application value in the secure training of sensitive data and disease prediction in the prototype system of the digital medical scenario.

Mingjie Chen. Research on Privacy-Preserving AI Model Training and Validation Methods Based on Federated Learning. International Journal of Multimedia Computing (2026), Vol. 7, Issue 1: 9-18. https://doi.org/10.38007/IJMC.2026.070102.

[1] Hayati H, Murguia C, Nathan V D W .Immersion and Invariance-based Coding for Privacy-Preserving Federated Learning [J]. 2024.

[2] Pan L, Huang M, Wang L, et al. FedDP: Privacy-preserving method based on federated learning for histopathology image segmentation [J]. 2024. DOI:10.1109/BIBM62325. 2024.10822021.

[3] Tang X, Wu X, Bao W. Intelligent Prediction-Inventory-Scheduling Closed-Loop Nearshore Supply Chain Decision System [J]. Advances in Management and Intelligent Technologies, 2025, 1(4).

[4] Chen H, Hsu H Y, Hsieh J Y, et al. A differential privacy-preserving federated learning scheme with predictive maintenance of wind turbines based on deep learning for feature compression and anomaly detection with state assessment [J].Journal of Mechanical Science & Technology, 2024, 38(7).DOI:10.1007/s12206-024-0616-9.

[5] Lu, Z. (2025). AI-Driven Cross-Cloud Operations Language Standardisation and Knowledge Sharing System. European Journal of AI, Computing & Informatics, 1(4), 43-50.

[6] Zhang, Xuanrui. "Automobile Finance Credit Fraud Risk Early Warning System based on Louvain Algorithm and XGBoost Model." In 2025 3rd International Conference on Data Science and Information System (ICDSIS), pp. 1-7. IEEE, 2025.

[7] Wu, H. (2025). The Commercialization Path of Large Language Models in Start-Ups. European Journal of Business, Economics & Management, 1(3), 38-44.

[8] Shen C, Zhang W, Zhou T, et al. An Efficient and Secure Privacy-Preserving Federated Learning Framework Based on Multiplicative Double Privacy Masking [J].Computers, Materials & Continua, 2024, 80(3):4729-4748.DOI:10.32604/cmc.2024.054434.

[9] Huang J, Chen Z, Liu S, et al. A Novel Federated Learning Framework Based on Conditional Generative Adversarial Networks for Privacy Preserving in 6G [J].Electronics, 2024, 13(4):20.DOI:10.3390/electronics13040783.

[10] Yu, X. (2025). Application Analysis of User Behavior Segmentation in Enhancing Customer Lifetime Value. Journal of Humanities, Arts and Social Science, 9(10).

[11] Chen J. Design and Implementation of a Personalized Recommendation System Based on Deep Learning Distributed Collaborative Filtering Algorithm on Social Media Platforms[C]//2025 3rd International Conference on Integrated Circuits and Communication Systems (ICICACS). IEEE, 2025: 1-5.

[12] Yang D, Liu X. Collaborative Algorithm for User Trust and Data Security Based on Blockchain and Machine Learning [J]. Procedia Computer Science, 2025, 262: 757-765.

[13] Lu, Z. (2025). Design and Practice of AI Intelligent Mentor System for DevOps Education. European Journal of Education Science, 1(3), 25-31.

[14] Zheng, H. (2025). Research on Delay-aware Scheduling Algorithms for Edge Task Migration in High-concurrency Environments. Engineering Advances, 5(4).

[15] Li, J. (2025). The Impact of Distributed Data Query Optimization on Large-Scale Data Processing.