.png){kind=icon}
返回
Distributed Boosting Variational Inference Algorithm Over Multi-Agent Networks
.png){kind=icon}
OA
AI
DOI：10.1109/ACCESS.2020.3033138.png)
.png){kind=icon}
.png){kind=icon}
分享.png){kind=icon}
收藏
摘要
En 中文
Distributed Bayesian estimation over multi-agent networks has received much attention due to its broad applications, where each agent has its private data that is unavailable to other agents. For efficient inference over multi-agent networks, we develop a distributed boosting variational inference (DBVI) algorithm with limited communication. We first decompose the global cost function into a sum-of-costs form, where each local cost only relates to its own dataset. Then, the global posterior distribution is approximated by a gradient decent at each boosting step, followed by a consensus protocol for cooperation with the neighbors. Moreover, we derive DBVI with Gaussian mixture model (DBVI-GMM) in detail. Finally, simulations on the synthetic and real datasets illustrate the effectiveness of the proposed algorithm.
Keyword:
Inference algorithms
Boosting
Approximation algorithms
Optimization
Gaussian mixture model
Bayes methods
Multi-agent networks
distributed machine learning
posterior probability approximation
boosting variational inference
AI总结
对已上传原文的论文进行重点信息的提取，主要内容包括：简要概述、研究摘要、背景介绍、关键亮点、图文解析、展望与总结。
期刊
IF:
3.6
论文数:
9.8W
被引数:
29.4W
机构
引用论文
分享收藏Diffusion-Based EM Algorithm for Distributed Estimation of Gaussian Mixtures in Wireless Sensor Networks.png)
SENSORSOAAI
SENSORSIF3.5
OAAI分享收藏Convergence analysis of distributed stochastic gradient descent with shuffling带shuffling的分布随机梯度下降算法的收敛性分析NEUROCOMPUTINGOAAI
NEUROCOMPUTINGIF6.5
OAAI分享收藏分享收藏
查看更多内容