arXiv subjects

Ali H. Sayed

Publications and source records attributed to Ali H. Sayed.

At least 91 records · Page 5Linked to original sources

arXiv2020

Graph-Homomorphic Perturbations for Private Decentralized Learning

Decentralized algorithms for stochastic optimization and learning rely on the diffusion of information as a result of repeated local exchanges of intermediate estimates. Such structures are particularly appealing in situations where agents may be hesitant to share raw data due to privacy concerns. Nevertheless, in the absence of additional privacy-preserving mechanisms, the exchange of local estimates, which are generated based on private data can allow for the inference of the data itself. The most common mechanism for guaranteeing privacy is the addition of perturbations to local estimates before broadcasting. These perturbations are generally chosen independently at every agent, resulting in a significant performance loss. We propose an alternative scheme, which constructs perturbations according to a particular nullspace condition, allowing them to be invisible (to first order in the step-size) to the network centroid, while preserving privacy guarantees. The analysis allows for general nonconvex loss functions, and is hence applicable to a large number of machine learning and signal processing problems, including deep learning.

cs.LG↗

arXiv2020

Dif-MAML: Decentralized Multi-Agent Meta-Learning

The objective of meta-learning is to exploit the knowledge obtained from observed tasks to improve adaptation to unseen tasks. As such, meta-learners are able to generalize better when they are trained with a larger number of observed tasks and with a larger amount of data per task. Given the amount of resources that are needed, it is generally difficult to expect the tasks, their respective data, and the necessary computational capacity to be available at a single central location. It is more natural to encounter situations where these resources are spread across several agents connected by some graph topology. The formalism of meta-learning is actually well-suited to this decentralized setting, where the learner would be able to benefit from information and computational power spread across the agents. Motivated by this observation, in this work, we propose a cooperative fully-decentralized multi-agent meta-learning algorithm, referred to as Diffusion-based MAML or Dif-MAML. Decentralized optimization algorithms are superior to centralized implementations in terms of scalability, avoidance of communication bottlenecks, and privacy guarantees. The work provides a detailed theoretical analysis to show that the proposed strategy allows a collection of agents to attain agreement at a linear rate and to converge to a stationary point of the aggregate MAML objective even in non-convex environments. Simulation results illustrate the theoretical findings and the superior performance relative to the traditional non-cooperative setting.

cs.LG↗

arXiv2020

Graph Learning Under Partial Observability

Many optimization, inference and learning tasks can be accomplished efficiently by means of decentralized processing algorithms where the network topology (i.e., the graph) plays a critical role in enabling the interactions among neighboring nodes. There is a large body of literature examining the effect of the graph structure on the performance of decentralized processing strategies. In this article, we examine the inverse problem and consider the reverse question: How much information does observing the behavior at the nodes of a graph convey about the underlying topology? For large-scale networks, the difficulty in addressing such inverse problems is compounded by the fact that usually only a limited fraction of the nodes can be probed, giving rise to a second important question: Despite the presence of unobserved nodes, can partial observations still be sufficient to discover the graph linking the probed nodes? The article surveys recent advances on this challenging learning problem and related questions.

cs.MA↗

arXiv2020

Decentralized Proximal Gradient Algorithms with Linear Convergence Rates

This work studies a class of non-smooth decentralized multi-agent optimization problems where the agents aim at minimizing a sum of local strongly-convex smooth components plus a common non-smooth term. We propose a general primal-dual algorithmic framework that unifies many existing state-of-the-art algorithms. We establish linear convergence of the proposed method to the exact solution in the presence of the non-smooth term. Moreover, for the more general class of problems with agent specific non-smooth terms, we show that linear convergence cannot be achieved (in the worst case) for the class of algorithms that uses the gradients and the proximal mappings of the smooth and non-smooth parts, respectively. We further provide a numerical counterexample that shows how some state-of-the-art algorithms fail to converge linearly for strongly-convex objectives and different local non-smooth terms.

math.OC↗

arXiv2020

A Multi-Agent Primal-Dual Strategy for Composite Optimization over Distributed Features

This work studies multi-agent sharing optimization problems with the objective function being the sum of smooth local functions plus a convex (possibly non-smooth) function coupling all agents. This scenario arises in many machine learning and engineering applications, such as regression over distributed features and resource allocation. We reformulate this problem into an equivalent saddle-point problem, which is amenable to decentralized solutions. We then propose a proximal primal-dual algorithm and establish its linear convergence to the optimal solution when the local functions are strongly-convex. To our knowledge, this is the first linearly convergent decentralized algorithm for multi-agent sharing problems with a general convex (possibly non-smooth) coupling function.

math.OC↗

arXiv2020

Graph Learning over Partially Observed Diffusion Networks: Role of Degree Concentration

This work examines the problem of graph learning over a diffusion network when data can be collected from a limited portion of the network (partial observability). The main question is to establish technical guarantees of consistent recovery of the subgraph of probed network nodes, i) despite the presence of unobserved nodes; and ii) under different connectivity regimes, including the dense regime where the probed nodes are influenced by many connections coming from the unobserved ones. We ascertain that suitable estimators of the combination matrix (i.e., the matrix that quantifies the pairwise interaction between nodes) possess an identifiability gap that enables the discrimination between connected and disconnected nodes. Fundamental conditions are established under which the subgraph of monitored nodes can be recovered, with high probability as the network size increases, through universal clustering algorithms. This claim is proved for three matrix estimators: i) the Granger estimator that adapts to the partial observability setting the solution that is exact under full observability ; ii) the one-lag correlation matrix; and iii) the residual estimator based on the difference between two consecutive time samples. A detailed characterization of the asymptotic behavior of these estimators is established in terms of an error bias and of the identifiability gap, and a sample complexity analysis is performed to establish how the number of samples scales with the network size to achieve consistent learning. Comparison among the estimators is performed through illustrative examples that show how estimators that are not optimal in the full observability regime can outperform the Granger estimator in the partial observability regime. The analysis reveals that the fundamental property enabling consistent graph learning is the statistical concentration of node degrees.

math.ST↗

arXiv2020

ISL: A novel approach for deep exploration

In this article we explore an alternative approach to address deep exploration and we introduce the ISL algorithm, which is efficient at performing deep exploration. Similarly to maximum entropy RL, we derive the algorithm by augmenting the traditional RL objective with a novel regularization term. A distinctive feature of our approach is that, as opposed to other works that tackle the problem of deep exploration, in our derivation both the learning equations and the exploration-exploitation strategy are derived in tandem as the solution to a well-posed optimization problem whose minimization leads to the optimal value function. Empirically we show that our method exhibits state of the art performance on a range of challenging deep-exploration benchmarks.

cs.LG↗

arXiv2020

Supervised Learning Under Distributed Features

This work studies the problem of learning under both large datasets and large-dimensional feature space scenarios. The feature information is assumed to be spread across agents in a network, where each agent observes some of the features. Through local cooperation, the agents are supposed to interact with each other to solve an inference problem and converge towards the global minimizer of an empirical risk. We study this problem exclusively in the primal domain, and propose new and effective distributed solutions with guaranteed convergence to the minimizer with linear rate under strong convexity. This is achieved by combining a dynamic diffusion construction, a pipeline strategy, and variance-reduced techniques. Simulation results illustrate the conclusions.

cs.MA↗

arXiv2020

Dynamic Federated Learning

Federated learning has emerged as an umbrella term for centralized coordination strategies in multi-agent environments. While many federated learning architectures process data in an online manner, and are hence adaptive by nature, most performance analyses assume static optimization problems and offer no guarantees in the presence of drifts in the problem solution or data characteristics. We consider a federated learning model where at every iteration, a random subset of available agents perform local updates based on their data. Under a non-stationary random walk model on the true minimizer for the aggregate optimization problem, we establish that the performance of the architecture is determined by three factors, namely, the data variability at each agent, the model variability across all agents, and a tracking term that is inversely proportional to the learning rate of the algorithm. The results clarify the trade-off between convergence and tracking performance.

cs.LG↗

arXiv2020

Diffusion LMS with Communication Delays: Stability and Performance Analysis

We study the problem of distributed estimation over adaptive networks where communication delays exist between nodes. In particular, we investigate the diffusion Least-Mean- Square (LMS) strategy where delayed intermediate estimates (due to the communication channels) are employed during the combination step. One important question is: Do the delays affect the stability condition and performance? To answer this question, we conduct a detailed performance analysis in the mean and in the mean-square-error sense of the diffusion LMS with delayed estimates. Stability conditions, transient and steady-state mean-square-deviation (MSD) expressions are provided. One of the main findings is that diffusion LMS with delays can still converge under the same step-sizes condition of the diffusion LMS without delays. Finally, simulation results illustrate the theoretical findings.

eess.SP↗

arXiv2020

Tracking Performance of Online Stochastic Learners

The utilization of online stochastic algorithms is popular in large-scale learning settings due to their ability to compute updates on the fly, without the need to store and process data in large batches. When a constant step-size is used, these algorithms also have the ability to adapt to drifts in problem parameters, such as data or model properties, and track the optimal solution with reasonable accuracy. Building on analogies with the study of adaptive filters, we establish a link between steady-state performance derived under stationarity assumptions and the tracking performance of online learners under random walk models. The link allows us to infer the tracking performance from steady-state expressions directly and almost by inspection.

math.OC↗

arXiv2020

Second-Order Guarantees in Centralized, Federated and Decentralized Nonconvex Optimization

Rapid advances in data collection and processing capabilities have allowed for the use of increasingly complex models that give rise to nonconvex optimization problems. These formulations, however, can be arbitrarily difficult to solve in general, in the sense that even simply verifying that a given point is a local minimum can be NP-hard [1]. Still, some relatively simple algorithms have been shown to lead to surprisingly good empirical results in many contexts of interest. Perhaps the most prominent example is the success of the backpropagation algorithm for training neural networks. Several recent works have pursued rigorous analytical justification for this phenomenon by studying the structure of the nonconvex optimization problems and establishing that simple algorithms, such as gradient descent and its variations, perform well in converging towards local minima and avoiding saddle-points. A key insight in these analyses is that gradient perturbations play a critical role in allowing local descent algorithms to efficiently distinguish desirable from undesirable stationary points and escape from the latter. In this article, we cover recent results on second-order guarantees for stochastic first-order optimization algorithms in centralized, federated, and decentralized architectures.

cs.MA↗

arXiv2020

Adaptation in Online Social Learning

This work studies social learning under non-stationary conditions. Although designed for online inference, classic social learning algorithms perform poorly under drifting conditions. To mitigate this drawback, we propose the Adaptive Social Learning (ASL) strategy. This strategy leverages an adaptive Bayesian update, where the adaptation degree can be modulated by tuning a suitable step-size parameter. The learning performance of the ASL algorithm is examined by means of a steady-state analysis. It is shown that, under the regime of small step-sizes: i) consistent learning is possible; ii) an accurate prediction of the performance can be furnished in terms of a Gaussian approximation.

eess.SP↗

arXiv2020

Learning Graph Influence from Social Interactions

In social learning, agents form their opinions or beliefs about certain hypotheses by exchanging local information. This work considers the recent paradigm of weak graphs, where the network is partitioned into sending and receiving components, with the former having the possibility of exerting a domineering effect on the latter. Such graph structures are prevalent over social platforms. We will not be focusing on the direct social learning problem (which examines what agents learn), but rather on the dual or reverse learning problem (which examines how agents learned). Specifically, from observations of the stream of beliefs at certain agents, we would like to examine whether it is possible to learn the strength of the connections (influences) from sending components in the network to these receiving agents.

cs.MA↗

arXiv2020

Affine Combination of Diffusion Strategies over Networks

Diffusion adaptation is a powerful strategy for distributed estimation and learning over networks. Motivated by the concept of combining adaptive filters, this work proposes a combination framework that aggregates the operation of multiple diffusion strategies for enhanced performance. By assigning a combination coefficient to each node, and using an adaptation mechanism to minimize the network error, we obtain a combined diffusion strategy that benefits from the best characteristics of all component strategies simultaneously in terms of excess-mean-square error (EMSE). Analyses of the universality are provided to show the superior performance of affine combination scheme and to characterize its behavior in the mean and mean-square sense. Simulation results are presented to demonstrate the effectiveness of the proposed strategies, as well as the accuracy of theoretical findings.

eess.SP↗

arXiv2020

Linear Convergence of Primal-Dual Gradient Methods and their Performance in Distributed Optimization

In this work, we revisit a classical incremental implementation of the primal-descent dual-ascent gradient method used for the solution of equality constrained optimization problems. We provide a short proof that establishes the linear (exponential) convergence of the algorithm for smooth strongly-convex cost functions and study its relation to the non-incremental implementation. We also study the effect of the augmented Lagrangian penalty term on the performance of distributed optimization algorithms for the minimization of aggregate cost functions over multi-agent networks.

math.OC↗

arXiv2019

Online Distributed Learning over Graphs with Multitask Graph-Filter Models

In this work, we are interested in adaptive and distributed estimation of graph filters from streaming data. We formulate this problem as a consensus estimation problem over graphs, which can be addressed with diffusion LMS strategies. Most popular graph-shift operators such as those based on the graph Laplacian matrix, or the adjacency matrix, are not energy preserving. This may result in an ill-conditioned estimation problem, and reduce the convergence speed of the distributed algorithms. To address this issue and improve the transient performance, we introduce a preconditioned graph diffusion LMS algorithm. We also propose a computationally efficient version of this algorithm by approximating the Hessian matrix with local information. Performance analyses in the mean and mean-square sense are provided. Finally, we consider a more general problem where the filter coefficients to estimate may vary over the graph. To avoid a large estimation bias, we introduce an unsupervised clustering method for splitting the global estimation problem into local ones. Numerical results show the effectiveness of the proposed algorithms and validate the theoretical results.

eess.SP↗

arXiv2019

Decision Learning and Adaptation over Multi-Task Networks

This paper studies the operation of multi-agent networks engaged in multi-task decision problems under the paradigm of simultaneous learning and adaptation. Two scenarios are considered: one in which a decision must be taken among multiple states of nature that are known but can vary over time and space, and another in which there exists a known "normal" state of nature and the task is to detect unpredictable and unknown deviations from it. In both cases the network learns from the past and adapts to changes in real time in a multi-task scenario with different clusters of agents addressing different decision problems. The system design takes care of challenging situations with clusters of complicated structure, and the performance assessment is conducted by computer simulations. A theoretical analysis is developed to obtain a statistical characterization of the agents' status at steady-state, under the simplifying assumption that clustering is made without errors. This provides approximate bounds for the steady-state decision performance of the agents. Insights are provided for deriving accurate performance prediction by exploiting the derived theoretical results.

eess.SP↗