📋 摘要
⭐ 联邦学习隐私与通信效率优化，与可信AI略有交集，但非SE/测试/形式化方法主线。
Enhanced Privacy and Communication Efficiency in Non-IID Federated Learning with Adaptive Quantization and Differential Privacy
中文
本文针对 Federated Learning (FL) 中通信瓶颈与隐私泄露两大挑战，提出将 differential privacy (DP) 与自适应量化相结合的训练方案。研究问题在于：跨设备带宽差异导致的通信开销，以及训练中通过模型或梯度分析暴露敏感信息的风险。方法上，作者采用 Laplacian-based DP 进行隐私保护，认为其相较于 FL 中常用的 Gaussian-based DP 探索较少且能提供更紧的隐私保证；同时提出两类比特长度调度器：基于 round-based cosine annealing 的全局调度器，以及依据 dataset entropy 评估客户端贡献并据此动态调整的 client-based 调度器。实验在 CIFAR10、MNIST 与医学影像数据集上展开，覆盖 non-IID 数据分布、不同客户端数量、比特调度策略以及多种隐私预算配置。结果显示，相较于 32-bit float 训练，所提自适应量化方法在 MNIST、CIFAR10 与医学影像数据集上分别最多降低 52.64%、45.06% 与 31%–37% 的通信数据量，同时保持具有竞争力的模型准确率并通过 DP 实现稳健隐私保护。与现有工作的差异主要体现在对 Laplacian DP 在 FL 场景的探索，以及结合全局与基于客户端贡献的双层自适应量化调度设计。
English abstract
Federated learning (FL) is a distributed machine learning method where multiple devices collaboratively train a model under the management of a central server without sharing underlying data. One of the key challenges of FL is the communication bottleneck caused by variations in connection speed and bandwidth across devices. Therefore, it is essential to reduce the size of transmitted data during training. Additionally, there is a potential risk of exposing sensitive information through the model or gradient analysis during training. To address both privacy and communication efficiency, we combine differential privacy (DP) and adaptive quantization methods. We use Laplacian-based DP to preserve privacy, which is relatively underexplored in FL and offers tighter privacy guarantees than Gaussian-based DP. We propose a simple and efficient global bit-length scheduler using round-based cosine annealing, along with a client-based scheduler that dynamically adapts based on client contribution estimated through dataset entropy analysis. We evaluate our approach through extensive experiments on CIFAR10, MNIST, and medical imaging datasets, using non-IID data distributions across varying client counts, bit-length schedulers, and privacy budgets. The results show that our adaptive quantization methods reduce total communicated data by up to 52.64% for MNIST, 45.06% for CIFAR10, and 31% to 37% for medical imaging datasets compared to 32-bit float training while maintaining competitive model accuracy and ensuring robust privacy through differential privacy.
加载中…
点文件 → 加为 tab；按 Esc 关闭