Distributed Pareto Optimization via Diffusion Strategies

Abstract

We consider solving multi-objective optimization problems in a distributed manner by a network of cooperating and learning agents. The problem is equivalent to optimizing a global cost that is the sum of individual components. The optimizers of the individual components do not necessarily coincide and the network therefore needs to seek Pareto optimal solutions. We develop a distributed solution that relies on a general class of adaptive diffusion strategies. We show how the diffusion process can be represented as the cascade composition of three operators: two combination operators and a gradient descent operator. Using the Banach fixed-point theorem, we establish the existence of a unique fixed point for the composite cascade. We then study how close each agent converges towards this fixed point, and also examine how close the Pareto solution is to the fixed point. We perform a detailed mean-square error analysis and establish that all agents are able to converge to the same Pareto optimal solution within a sufficiently small mean-square-error (MSE) bound even for constant step-sizes. We illustrate one application of the theory to collaborative decision making in finance by a network of agents.

35 pages, 9 figures, submitted for publication

Cited by in corpus (42)

• Multitask Diffusion Adaptation over Networks
• Diffusion LMS over Multitask Networks
• Distributed Clustering and Learning Over Networks
• Adaptive Penalty-Based Distributed Stochastic Convex Optimization
• Dictionary Learning over Distributed Models
• Proximal Multitask Learning over Networks with Sparsity-inducing Coregularization
• Study of Robust Diffusion Recursive Least Squares Algorithms with Side Information for Networked Agents
• Empirical Centroid Fictitious Play: An Approach For Distributed Learning In Multi-Agent Games
• Diffusion LMS for Multitask Problems with Local Linear Equality Constraints
• On the Influence of Bias-Correction on Distributed Stochastic Optimization
• Multitask diffusion adaptation over networks with common latent representations
• Supervised Learning Under Distributed Features
• Diffusion-Based Adaptive Distributed Detection: Steady-State Performance in the Slow Adaptation Regime
• A Unified and Refined Convergence Analysis for Non-Convex Decentralized Learning
• A Linearly Convergent Proximal Gradient Algorithm for Decentralized Optimization
• On reducing the communication cost of the diffusion LMS algorithm
• Energy Efficiency of Distributed Signal Processing in Wireless Networks: A Cross-Layer Analysis
• Can Primal Methods Outperform Primal-dual Methods in Decentralized Dynamic Optimization?
• Diff-DAC: Distributed Actor-Critic for Average Multitask Deep Reinforcement Learning
• Diffusion Leaky Zero Attracting Least Mean Square Algorithm and Its Performance Analysis
• Information-Sharing over Adaptive Networks with Self-interested Agents
• Exact Diffusion for Distributed Optimization and Learning --- Part I: Algorithm Development
• Distributed Adaptive Learning Under Communication Constraints
• Removing Data Heterogeneity Influence Enhances Network Topology Dependence of Decentralized SGD
• Decentralized Consensus Optimization with Asynchrony and Delays
• Tracking Performance of Online Stochastic Learners
• Compressed Regression over Adaptive Networks
• Asynchronous Adaptation and Learning over Networks - Part II: Performance Analysis
• Asynchronous Adaptation and Learning over Networks --- Part I: Modeling and Stability Analysis
• Asynchronous adaptive networks
• Characterization and Control of Diffusion Processes in Multi-Agent Networks
• Information Exchange and Learning Dynamics over Weakly-Connected Adaptive Networks
• Fully Distributed Actor-Critic Architecture for Multitask Deep Reinforcement Learning
• Stochastic Subgradient Algorithms for Strongly Convex Optimization over Distributed Networks
• Noise-Robust and Resource-Efficient ADMM-based Federated Learning
• Dynamic Average Diffusion with randomized Coordinate Updates
• Study of Robust Distributed Diffusion RLS Algorithms with Side Information for Adaptive Networks
• On the Asymptotic Bias of the Diffusion-Based Distributed Pareto Optimization
• Multi-Agent Optimization and Learning: A Non-Expansive Operators Perspective
• Study of Diffusion Normalized Least Mean M-estimate Algorithms
• Diffusion LMS for clustered multitask networks
• Privacy-Preserving Distributed Projection LMS for Linear Multitask Networks