arXiv subjects

Nicola Bastianello

Publications and source records attributed to Nicola Bastianello.

At least 37 records · Page 2Linked to original sources

arXiv2025

Control-Based Online Distributed Optimization

In this paper we design a novel class of online distributed optimization algorithms leveraging control theoretical techniques. We start by focusing on quadratic costs, and assuming to know an internal model of their variation. In this set-up, we formulate the algorithm design as a robust control problem, showing that it yields a fully distributed algorithm. We also provide a distributed routine to acquire the internal model. We show that the algorithm converges exactly to the sequence of optimal solutions. We empirically evaluate the performance of the algorithm for different choices of parameters. Additionally, we evaluate the performance of the algorithm for quadratic problems with inexact internal model and non-quadratic problems, and show that it outperforms alternative algorithms in both scenarios.

math.OC↗

arXiv2025

Distributed Optimization and Learning for Automated Stepsize Selection with Finite Time Coordination

Distributed optimization and learning algorithms are designed to operate over large scale networks enabling processing of vast amounts of data effectively and efficiently. One of the main challenges for ensuring a smooth learning process in gradient-based methods is the appropriate selection of a learning stepsize. Most current distributed approaches let individual nodes adapt their stepsizes locally. However, this may introduce stepsize heterogeneity in the network, thus disrupting the learning process and potentially leading to divergence. In this paper, we propose a distributed learning algorithm that incorporates a novel mechanism for automating stepsize selection among nodes. Our main idea relies on implementing a finite time coordination algorithm for eliminating stepsize heterogeneity among nodes. We analyze the operation of our algorithm and we establish its convergence to the optimal solution. We conclude our paper with numerical simulations for a linear regression problem, showcasing that eliminating stepsize heterogeneity enhances convergence speed and accuracy against current approaches.

eess.SY↗

arXiv2025

Modular Distributed Nonconvex Learning with Error Feedback

In this paper, we design a novel distributed learning algorithm using stochastic compressed communications. In detail, we pursue a modular approach, merging ADMM and a gradient-based approach, benefiting from the robustness of the former and the computational efficiency of the latter. Additionally, we integrate a stochastic integral action (error feedback) enabling almost sure rejection of the compression error. We analyze the resulting method in nonconvex scenarios and guarantee almost sure asymptotic convergence to the set of stationary points of the problem. This result is obtained using system-theoretic tools based on stochastic timescale separation. We corroborate our findings with numerical simulations in nonconvex classification.

math.OC↗

arXiv2025

(Un)supervised Learning of Maximal Lyapunov Functions

In this paper, we address the problem of discovering maximal Lyapunov functions, as a means of determining the region of attraction of a dynamical system. To this end, we design a novel neural network architecture, which we prove to be a universal approximator of (maximal) Lyapunov functions. The architecture combines a local quadratic approximation with the output of a neural network, which models global higher-order terms in the Taylor expansion. We formulate the problem of training the Lyapunov function as an unsupervised optimization problem with dynamical constraints, which can be solved leveraging techniques from physics-informed learning. We propose and analyze a tailored training algorithm, based on the primal-dual algorithm, that can efficiently solve the problem. Additionally, we show how the learning problem formulation can be adapted to integrate data, when available. We apply the proposed approach to different classes of systems, showing that it matches or outperforms state-of-the-art alternatives in the accuracy of the approximated regions of attraction.

math.OC↗

arXiv2025

ADMM-Tracking Gradient for Distributed Optimization over Asynchronous and Unreliable Networks

In this paper, we propose a novel distributed algorithm for consensus optimization over networks and a robust extension tailored to deal with asynchronous agents and packet losses. Indeed, to robustly achieve dynamic consensus on the solution estimates and the global descent direction, we embed in our algorithms a distributed implementation of the Alternating Direction Method of Multipliers (ADMM). Such a mechanism is suitably interlaced with a local proportional action steering each agent estimate to the solution of the original consensus optimization problem. First, in the case of ideal networks, by using tools from system theory, we prove the linear convergence of the scheme with strongly convex costs. Then, by exploiting the averaging theory, we extend such a first result to prove that the robust extension of our method preserves linear convergence in the case of asynchronous agents and packet losses. Further, by using the notion of Input-to-State Stability, we also guarantee the robustness of the schemes with respect to additional, generic errors affecting the agents' updates. Finally, some numerical simulations confirm our theoretical findings and compare our algorithms with other distributed schemes in terms of speed and robustness.

math.OC↗

arXiv2025

Optimization and Learning in Open Multi-Agent Systems

Modern artificial intelligence relies on networks of agents that collect data, process information, and exchange it with neighbors to collaboratively solve optimization and learning problems. This article introduces a novel distributed algorithm to address a broad class of these problems in "open networks", where the number of participating agents may vary due to several factors, such as autonomous decisions, heterogeneous resource availability, or DoS attacks. Extending the current literature, the convergence analysis of the proposed algorithm is based on the newly developed "Theory of Open Operators", which characterizes an operator as open when the set of components to be updated changes over time, yielding to time-varying operators acting on sequences of points of different dimensions and compositions. The mathematical tools and convergence results developed here provide a general framework for evaluating distributed algorithms in open networks, allowing to characterize their performance in terms of the punctual distance from the optimal solution, in contrast with regret-based metrics that assess cumulative performance over a finite-time horizon. As illustrative examples, the proposed algorithm is used to solve dynamic consensus or tracking problems on different metrics of interest, such as average, median, and min/max value, as well as classification problems with logistic loss functions.

math.OC↗

arXiv2024

Enhancing Privacy in Federated Learning through Local Training

In this paper we propose the federated learning algorithm Fed-PLT to overcome the challenges of (i) expensive communications and (ii) privacy preservation. We address (i) by allowing for both partial participation and local training, which significantly reduce the number of communication rounds between the central coordinator and computing agents. The algorithm matches the state of the art in the sense that the use of local training demonstrably does not impact accuracy. Additionally, agents have the flexibility to choose from various local training solvers, such as (stochastic) gradient descent and accelerated gradient descent. Further, we investigate how employing local training can enhance privacy, addressing point (ii). In particular, we derive differential privacy bounds and highlight their dependence on the number of local training epochs. We assess the effectiveness of the proposed algorithm by comparing it to alternative techniques, considering both theoretical analysis and numerical results from a classification task.

cs.LG↗

arXiv2024

A Control Theoretical Approach to Online Constrained Optimization

In this paper we focus on the solution of online problems with time-varying, linear equality and inequality constraints. Our approach is to design a novel online algorithm by leveraging the tools of control theory. In particular, for the case of equality constraints only, using robust control we design an online algorithm with asymptotic convergence to the optimal trajectory, differently from the alternatives that achieve non-zero tracking error. When also inequality constraints are present, we show how to modify the proposed algorithm to account for the wind-up induced by the nonnegativity constraints on the dual variables. We report numerical results that corroborate the theoretical analysis, and show how the proposed approach outperforms state-of-the-art algorithms both with equality and inequality constraints.

math.OC↗

arXiv2024

Hierarchical Federated ADMM

In this paper, we depart from the widely-used gradient descent-based hierarchical federated learning (FL) algorithms to develop a novel hierarchical FL framework based on the alternating direction method of multipliers (ADMM). Within this framework, we propose two novel FL algorithms, which both use ADMM in the top layer: one that employs ADMM in the lower layer and another that uses the conventional gradient descent-based approach. The proposed framework enhances privacy, and experiments demonstrate the superiority of the proposed algorithms compared to the conventional algorithms in terms of learning convergence and accuracy. Additionally, gradient descent on the lower layer performs well even if the number of local steps is very limited, while ADMM on both layers lead to better performance otherwise.

cs.LG↗

arXiv2024

Asynchronous Distributed Learning with Quantized Finite-Time Coordination

In this paper we address distributed learning problems over peer-to-peer networks. In particular, we focus on the challenges of quantized communications, asynchrony, and stochastic gradients that arise in this set-up. We first discuss how to turn the presence of quantized communications into an advantage, by resorting to a finite-time, quantized coordination scheme. This scheme is combined with a distributed gradient descent method to derive the proposed algorithm. Secondly, we show how this algorithm can be adapted to allow asynchronous operations of the agents, as well as the use of stochastic gradients. Finally, we propose a variant of the algorithm which employs zooming-in quantization. We analyze the convergence of the proposed methods and compare them to state-of-the-art alternatives.

math.OC↗

arXiv2024

A survey on secure decentralized optimization and learning

Decentralized optimization has become a standard paradigm for solving large-scale decision-making problems and training large machine learning models without centralizing data. However, this paradigm introduces new privacy and security risks, with malicious agents potentially able to infer private data or impair the model accuracy. Over the past decade, significant advancements have been made in developing secure decentralized optimization and learning frameworks and algorithms. This survey provides a comprehensive tutorial on these advancements. We begin with the fundamentals of decentralized optimization and learning, highlighting centralized aggregation and distributed consensus as key modules exposed to security risks in federated and distributed optimization, respectively. Next, we focus on privacy-preserving algorithms, detailing three cryptographic tools and their integration into decentralized optimization and learning systems. Additionally, we examine resilient algorithms, exploring the design and analysis of resilient aggregation and consensus protocols that support these systems. We conclude the survey by discussing current trends and potential future directions.

cs.LG↗

arXiv2024

A Stochastic Operator Framework for Optimization and Learning with Sub-Weibull Errors

This paper proposes a framework to study the convergence of stochastic optimization and learning algorithms. The framework is modeled over the different challenges that these algorithms pose, such as (i) the presence of random additive errors (e.g. due to stochastic gradients), and (ii) random coordinate updates (e.g. due to asynchrony in distributed set-ups). The paper covers both convex and strongly convex problems, and it also analyzes online scenarios, involving changes in the data and costs. The paper relies on interpreting stochastic algorithms as the iterated application of stochastic operators, thus allowing us to use the powerful tools of operator theory. In particular, we consider operators characterized by additive errors with sub-Weibull distribution (which parameterize a broad class of errors by their tail probability), and random updates. In this framework we derive convergence results in mean and in high probability, by providing bounds to the distance of the current iteration from a solution of the optimization or learning problem. The contributions are discussed in light of federated learning applications.

math.OC↗

arXiv2024

Multi-Agent Optimization and Learning: A Non-Expansive Operators Perspective

Multi-agent systems are increasingly widespread in a range of application domains, with optimization and learning underpinning many of the tasks that arise in this context. Different approaches have been proposed to enable the cooperative solution of these optimization and learning problems, including first- and second-order methods, and dual (or Lagrangian) methods, all of which rely on consensus and message-passing. In this article we discuss these algorithms through the lens of non-expansive operator theory, providing a unifying perspective. We highlight the insights that this viewpoint delivers, and discuss how it can spark future original research.

math.OC↗

arXiv2024

Robust Online Learning over Networks

The recent deployment of multi-agent networks has enabled the distributed solution of learning problems, where agents cooperate to train a global model without sharing their local, private data. This work specifically targets some prevalent challenges inherent to distributed learning: (i) online training, i.e., the local data change over time; (ii) asynchronous agent computations; (iii) unreliable and limited communications; and (iv) inexact local computations. To tackle these challenges, we apply the Distributed Operator Theoretical (DOT) version of the Alternating Direction Method of Multipliers (ADMM), which we call "DOT-ADMM". We prove that if the DOT-ADMM operator is metric subregular, then it converges with a linear rate for a large class of (not necessarily strongly) convex learning problems toward a bounded neighborhood of the optimal time-varying solution, and characterize how such neighborhood depends on (i)-(iv). We first derive an easy-to-verify condition for ensuring the metric subregularity of an operator, followed by tutorial examples on linear and logistic regression problems. We corroborate the theoretical analysis with numerical simulations comparing DOT-ADMM with other state-of-the-art algorithms, showing that only the proposed algorithm exhibits robustness to (i)-(iv).

math.OC↗

arXiv2024

Energy Reduction in Cell-Free Massive MIMO through Fine-Grained Resource Management

The physical layer foundations of cell-free massive MIMO (CF-mMIMO) have been well-established. As a next step, researchers are investigating practical and energy-efficient network implementations. This paper focuses on multiple sets of access points (APs) where user equipments (UEs) are served in each set, termed a federation, without inter-federation interference. The combination of federations and CF-mMIMO shows promise for highly-loaded scenarios. Our aim is to minimize the total energy consumption while adhering to UE downlink data rate constraints. The energy expenditure of the full system is modelled using a detailed hardware model of the APs. We jointly design the AP-UE association variables, determine active APs, and assign APs and UEs to federations. To solve this highly combinatorial problem, we develop a novel alternating optimization algorithm. Simulation results for an indoor factory demonstrate the advantages of considering multiple federations, particularly when facing large data rate requirements. Furthermore, we show that adopting a more distributed CF-mMIMO architecture is necessary to meet the data rate requirements. Conversely, if feasible, using a less distributed system with more antennas at each AP is more advantageous from an energy savings perspective.

eess.SP↗

arXiv2023

Online Distributed Learning with Quantized Finite-Time Coordination

In this paper we consider online distributed learning problems. Online distributed learning refers to the process of training learning models on distributed data sources. In our setting a set of agents need to cooperatively train a learning model from streaming data. Differently from federated learning, the proposed approach does not rely on a central server but only on peer-to-peer communications among the agents. This approach is often used in scenarios where data cannot be moved to a centralized location due to privacy, security, or cost reasons. In order to overcome the absence of a central server, we propose a distributed algorithm that relies on a quantized, finite-time coordination protocol to aggregate the locally trained models. Furthermore, our algorithm allows for the use of stochastic gradients during local training. Stochastic gradients are computed using a randomly sampled subset of the local training data, which makes the proposed algorithm more efficient and scalable than traditional gradient descent. In our paper, we analyze the performance of the proposed algorithm in terms of the mean distance from the online solution. Finally, we present numerical results for a logistic regression task.

cs.LG↗

arXiv2023

Feedback-Based Optimization with Sub-Weibull Gradient Errors and Intermittent Updates

This paper considers a feedback-based projected gradient method for optimizing systems modeled as algebraic maps. The focus is on a setup where the gradient is corrupted by random errors that follow a sub-Weibull distribution, and where the measurements of the output -- which replace the input-output map of the system in the algorithmic updates -- may not be available at each iteration. The sub-Weibull error model is particularly well-suited in frameworks where the cost of the problem is learned via Gaussian Process (GP) regression (from functional evaluations) concurrently with the execution of the algorithm; however, it also naturally models setups where nonparametric methods and neural networks are utilized to estimate the cost. Using the sub-Weibull model, and with Bernoulli random variables modeling missing measurements of the system output, we show that the online algorithm generates points that are within a bounded error from the optimal solutions. In particular, we provide error bounds in expectation and in high probability. Numerical results are presented in the context of a demand response problem in smart power grids.

math.OC↗

arXiv2023

Extrapolation-based Prediction-Correction Methods for Time-varying Convex Optimization

In this paper, we focus on the solution of online optimization problems that arise often in signal processing and machine learning, in which we have access to streaming sources of data. We discuss algorithms for online optimization based on the prediction-correction paradigm, both in the primal and dual space. In particular, we leverage the typical regularized least-squares structure appearing in many signal processing problems to propose a novel and tailored prediction strategy, which we call extrapolation-based. By using tools from operator theory, we then analyze the convergence of the proposed methods as applied both to primal and dual problems, deriving an explicit bound for the tracking error, that is, the distance from the time-varying optimal solution. We further discuss the empirical performance of the algorithm when applied to signal processing, machine learning, and robotics problems.

math.OC↗