میز خدمت الکترونیک
دوره ویراستاری

  اخبار و اعلانات

 آمار

تعداد نشریات127
تعداد شماره‌ها7,140
تعداد مقالات76,843
تعداد مشاهده مقاله154,469,522
تعداد دریافت فایل اصل مقاله116,530,123
Asadi, Arvin, Saadat, Mahsa. (1404). ADMM-DP: A Distributed and Privacy-Preserving Optimization Framework for Scalable Machine Learning in Information Systems. , 57(2), 25-65. doi: 10.22059/jac.2025.404651.1245
Arvin Asadi; Mahsa Saadat. "ADMM-DP: A Distributed and Privacy-Preserving Optimization Framework for Scalable Machine Learning in Information Systems". , 57, 2, 1404, 25-65. doi: 10.22059/jac.2025.404651.1245
Asadi, Arvin, Saadat, Mahsa. (1404). 'ADMM-DP: A Distributed and Privacy-Preserving Optimization Framework for Scalable Machine Learning in Information Systems', , 57(2), pp. 25-65. doi: 10.22059/jac.2025.404651.1245
Asadi, Arvin, Saadat, Mahsa. ADMM-DP: A Distributed and Privacy-Preserving Optimization Framework for Scalable Machine Learning in Information Systems. , 1404; 57(2): 25-65. doi: 10.22059/jac.2025.404651.1245

ADMM-DP: A Distributed and Privacy-Preserving Optimization Framework for Scalable Machine Learning in Information Systems

Journal of Algorithms and Computation
دوره 57، شماره 2، اسفند 2025، صفحه 25-65    اصل مقاله (1.65 M)
نوع مقاله: Research Paper
شناسه دیجیتال (DOI): 10.22059/jac.2025.404651.1245
نویسندگان
Arvin Asadi* 1؛ Mahsa Saadat2
1Department of Mathematical Sciences Sharif University of Technology, Tehran, Iran
2Department of Mathematics and Computer Science, Amirkabir University of Technology, Tehran, Iran
چکیده
Recent advances in federated learning and IoT-driven edge analytics underscore the need for optimization techniques that are both scalable and privacy-preserving[1][2]. In this work, we introduce ADMM-DP, a variant of the Alternating Direction Method of Multipliers that seamlessly integrates differential privacy (DP) guarantees in a fully decentralized, multi-agent learning architecture[3]. ADMM-DP leverages an augmented Lagrangian formulation with adaptive inexact local updates and calibrated Gaussian noise injection into each exchanged message, ensuring rigorous (ε,δ)-DP without sacrificing convergence[4][5]. Theoretically, we establish convergence rates and privacy-utility bounds under realistic heterogeneous (non-IID) data conditions. Building on DP-ADMM literature, we prove that ADMM-DP converges to a stationary solution with an explicit utility-privacy tradeoff[6], and furthermore, for strongly convex losses the method attains linear convergence rates comparable to non-private ADMM[7]. Privacy loss is tracked via advanced composition (moments accountant) to yield tight end-to-end DP guarantees[8].
کلیدواژه‌ها
Big data preprocessing؛ Machine learning؛ Feature selection؛ Data mining؛ self optimizing data؛ Apache Spark؛ Flink؛ Ray
آمار
تعداد مشاهده مقاله: 89
تعداد دریافت فایل اصل مقاله: 61





سامانه مدیریت نشریات علمی. قدرت گرفته از سیناوب