Privacy-Aware Collaborative and Distributed Bayesian Optimization

2026-07-13

AI Summary

This work addresses the privacy risks in distributed Bayesian optimization arising from gradient sharing, which may inadvertently leak clients’ sensitive observations. To mitigate this issue, the authors propose a privacy-preserving collaborative meta-learning framework that avoids exchanging raw data and instead integrates differential privacy mechanisms during the later stages of optimization to effectively reduce gradient-induced privacy leakage. The approach explicitly quantifies the trade-off between privacy guarantees and model utility. Experimental results demonstrate that the proposed method achieves performance comparable to centralized Bayesian optimization while providing strong privacy protection, and systematically elucidates the inherent privacy-utility trade-off under varying privacy budgets.

Abstract

We propose a collaborative meta-learning framework for distributed Bayesian optimization matching centralized performance without raw-data exchange. We show gradient sharing leaks client observations, with leakage worsening as the search converges and queries concentrate near the optimum. We evaluate a differentially private defense and characterize its privacy-utility trade-off.

Problem

• privacy leakage
• distributed Bayesian optimization
• gradient sharing
• differential privacy
• collaborative learning

Innovation

• Distributed Bayesian Optimization
• Privacy Leakage
• Gradient Sharing
• Differential Privacy
• Meta-Learning