突破（决策）边界：在联邦学习中动态校准差分隐私噪声以适配可解释性
Pushing the (Decision) Boundaries: Dynamically Calibrating Differentially Private Noise to Explainability in Federated Learning
浏览论文内容
中文总结 AI 辅助
提出XCal-FL算法，在跨筒仓联邦学习中从三个信号动态校准差分隐私噪声，提升模型准确性与解释保真度，实现更高隐私预算效率，为决策关键应用提供支持。
中文摘要 AI 辅助
结合差分隐私（DP）的联邦学习（FL）被越来越多地用于保护分布式机器学习中的数据机密性，但DP噪声会扭曲学习到的表示并降低解释保真度，这限制了需要可信解释的差分隐私FL应用，如辅助临床诊断。现有工作采用静态特征重要性信号调整DP噪声，将可解释性限制为事后分析，无法在训练期间将噪声校准至解释质量。我们提出XCal-FL，这是一种用于跨筒仓FL图像分类的闭环、可解释性驱动的本地训练算法，它从三个互补信号动态校准DP噪声：（1）预测对数几率变化，衡量对模型置信度的因果影响；（2）反事实间隔，捕捉决策边界敏感性；（3）显著性集中度，量化模型注意力的空间一致性，同时通过自适应隐私核算确保形式化DP保证。在三个医学成像数据集上针对不同FL配置开展的实验表明，XCal-FL生成的全局模型更准确且可解释，与静态噪声FL相比，预测性能提升超10%，解释保真度提升最高达5倍，且在保真度上优于最先进的自适应DP方法。XCal-FL还实现了更高的隐私预算效率，使每单位累积隐私损失能在准确性和解释保真度上获得更大增益。我们的分析进一步揭示，与随隐私损失大致线性缩放的预测性能不同，解释保真度呈现非线性动态，这些发现表明可解释性是隐私权衡中一个独特的维度，无法仅从效用推断，对决策关键应用的训练和隐私预算分配具有重要意义。
英文摘要
Federated Learning (FL) with Differential Privacy (DP) is increasingly adopted to preserve data confidentiality in distributed machine learning. However, DP noise distorts learned representations and degrades explanation fidelity, limiting differentially private FL where trustworthy explanations are required, such as assistive clinical diagnosis. Prior work adapted DP noise with static feature-importance signals, restricting explainability to post hoc analysis and precluding noise calibration to explanation quality during training. We propose XCal-FL, a closed-loop, explainability-driven local training algorithm for image classification in cross-silo FL that dynamically calibrates DP noise from three complementary signals: (1) prediction logit variations, measuring causal influence on model confidence, (2) counterfactual margins, capturing decision-boundary sensitivity, and (3) saliency concentration, quantifying spatial coherence of model attention, while enforcing formal DP guarantees via adaptive privacy accounting. Experiments on three medical imaging datasets across varying FL configurations show that XCal-FL yields more accurate and interpretable global models, improving predictive performance by over 10\% and explanation fidelity by up to 5$\times$ over static-noise FL, and outperforming state-of-the-art adaptive DP methods in fidelity. XCal-FL also achieves higher privacy-budget efficiency, turning each unit of cumulative privacy loss into larger gains in both accuracy and explanation fidelity. Our analysis further reveals that, unlike predictive performance, which scales roughly linearly with privacy loss, explanation fidelity exhibits non-linear dynamics. These findings suggest explainability is a distinct dimension of the privacy trade-off that cannot be inferred from utility alone, with implications for training and privacy-budget allocation in decision-critical applications.
发表机构
- School of Industrial & Intelligent Systems Engineering, Tel Aviv University（特拉维夫大学工业与智能系统工程学院）
- School of Integrated Technology, Yonsei University（延世大学融合技术学院）
机构由 AI 辅助整理，请以论文原文为准。
