Recycled ADMM: Improve Privacy and Accuracy with Less Computation in Distributed Algorithms
Authors:
- 1 October 2018
- conference paper
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- p. 959-965
Abstract
Alternating direction method of multiplier (ADMM) is a powerful method to solve decentralized convex optimization problems. In distributed settings, each node performs computation with its local data and the local results are exchanged among neighboring nodes in an iterative fashion. During this iterative process the leakage of data privacy arises and can accumulate significantly over many iterations, making it difficult to balance the privacy-utility tradeoff. In this study we propose Recycled ADMM (R-ADMM), where a linear approximation is applied to every even iteration, its solution directly calculated using only results from the previous, odd iteration. It turns out that under such a scheme, half of the updates incur no privacy loss and require much less computation compared to the conventional ADMM. We obtain a sufficient condition for the convergence of R-ADMM and provide the privacy analysis based on objective perturbation.Keywords
All Related Versions
This publication has 21 references indexed in Scilit:
- Privacy-Preserving Data Aggregation over Incomplete Data for Crowdsensing2016
- Dynamic Differential Privacy for ADMM-Based Distributed Classification LearningIEEE Transactions on Information Forensics and Security, 2016
- Weighted ADMM for Fast Decentralized Network OptimizationIEEE Transactions on Signal Processing, 2016
- Differentially Private Distributed Constrained OptimizationIEEE Transactions on Automatic Control, 2016
- Decentralized quadratically approximated alternating direction method of multipliers2015
- Differentially private cloud-based multi-agent optimization with constraints2015
- DLM: Decentralized Linearized Alternating Direction Method of MultipliersIEEE Transactions on Signal Processing, 2015
- Differentially Private Distributed Optimization2015
- Decentralized linearized alternating direction method of multipliers2014
- Distributed Alternating Direction Method of Multipliers2012