| Title | Recycled ADMM: Improve Privacy and Accuracy with Less Computation in Distributed Algorithms |
| Publication Type | Conference Paper |
| Year of Publication | 2018 |
| Authors | Zhang, Xueru, Khalili, Mohammad Mahdi, Liu, Mingyan |
| Conference Name | 2018 56th Annual Allerton Conference on Communication, Control, and Computing (Allerton) |
| Date Published | oct |
| Keywords | alternating direction method of multiplier, approximation theory, Computing Theory and Privacy, convergence, Convex functions, convex programming, data privacy, decentralized convex optimization problems, Differential privacy, distributed algorithms, Human Behavior, Iterative methods, iterative process, linear approximation, local data, Optimization, Perturbation methods, privacy, privacy analysis, privacy-utility tradeoff, pubcrawl, R-ADMM, recycled ADMM, Resiliency, Scalability |
| 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. |
| DOI | 10.1109/ALLERTON.2018.8635916 |
| Citation Key | zhang_recycled_2018 |
Recycled ADMM: Improve Privacy and Accuracy with Less Computation in Distributed Algorithms