arXiv cs.AI· Alexander Gr\"afe, Max van Gemmeren, Paul Brunzema, Sebastian Trimpe·· 1 天前AI 评分24
GLOBAFED：用神经网络代理实现联邦贝叶斯优化中的高效知识迁移
Efficient Knowledge Transfer in Federated Bayesian Optimization through Neural Network Surrogates
AI 导读
GLOBAFED 框架通过带共享特征提取主干和智能体专属头的概率神经网络代理，在联邦贝叶斯优化中实现不同但相关目标间的知识迁移，无需暴露原始数据。智能体在共享主干上做联邦学习构建共同潜在特征表示，专属头适配各自目标，并借助变分贝叶斯末层给出不确定性估计。实验显示其收敛速度和最终性能均优于现有 FedBO 基线，还能用已学到的全局特征热启动优化。
正文 · 原文
Abstract:Federated Bayesian Optimization (FedBO) has emerged as a powerful paradigm for enhancing Bayesian Optimization (BO) in distributed, privacy-sensitive settings, enabling agents to accelerate the optimization of their local objectives by sharing knowledge about their similarities without exposing raw data. However, existing methods rely on restrictive assumptions, such as near-identical objective functions or shared Gaussian process hyperparameters that are often violated in heterogeneous real-world applications. We present GLOBAFED, a novel framework that enables efficient knowledge transfer across different but related objectives without compromising privacy. At its core is a probabilistic neural network surrogate with a shared feature-extracting backbone and agent-specific heads equipped with varia- tional Bayesian last layers to provide uncertainty estimates. During optimization, agents use federated learning on the shared backbone to collaboratively build a common latent feature representation, while agent-specific heads adapt to their individual objectives. Our empirical results demonstrate that GLOBAFED significantly accelerates convergence and reaches better final performance compared to state-of-the-art FedBO baselines, and enables agents to warm-start their optimization from previously learned global features.
| Subjects: | Machine Learning (cs.LG); Artificial Intelligence (cs.AI) |
| Cite as: | arXiv:2611.15707 [cs.LG] |
| (or arXiv:2611.15707v1 [cs.LG] for this version) | |
| https://doi.org/10.48550/arXiv.2611.15707 arXiv-issued DOI via DataCite (pending registration) |
Submission history
From: Alexander Gräfe [view email]
[v1]
Wed, 11 Nov 2026 16:48:55 UTC (1,429 KB)
来源：arXiv cs.AI · arxiv.org