Layer Selection for Secure Federated Learning in Wireless Large AI Model (WLAM)
Yahao Ding, Zhaohui Yang, Mohammad Shikh‐Bahaei · 2024
In this article, the problem of training secure federated learning (FL) algorithm over a multi-cell wireless network for wireless large AI model (WLAM) is investigated. FL is indeed a learning method that can protect users’ privacy, but it has also been shown to be vulnerable to gradient leakage attacks, which can leak users’ private data. Therefore, we propose a defense method to prevent gradient leakage attacks by uploading the selected layers to contribute to global model updates. Moreover, we apply the differential privacy (DP) method to strengthen the defense during the local training process. We formulate an optimization problem that minimizes this leakage by jointly optimizing resource block (RB) allocation and layer selection. The numerical results indicate that the proposed algorithm can effectively reduce privacy leakage under communication constraints.