arXiv subjects

Rohit Parasnis

Publications and source records attributed to Rohit Parasnis.

11 recordsLinked to original sources

arXiv2025

Adversarial Node Placement in Decentralized Federated Learning: Maximum Spanning-Centrality Strategy and Performance Analysis

As Federated Learning (FL) becomes more widespread, there is growing interest in its decentralized variants. Decentralized FL leverages the benefits of fast and energy-efficient device-to-device communications to obviate the need for a central server. However, this opens the door to new security vulnerabilities as well. While FL security has been a popular research topic, the role of adversarial node placement in decentralized FL remains largely unexplored. This paper addresses this gap by evaluating the impact of various coordinated adversarial node placement strategies on decentralized FL's model training performance. We adapt two threads of placement strategies to this context: maximum span-based algorithms, and network centrality-based approaches. Building on them, we propose a novel attack strategy, MaxSpAN-FL, which is a hybrid between these paradigms that adjusts node placement probabilistically based on network topology characteristics. Numerical experiments demonstrate that our attack consistently induces the largest degradation in decentralized FL models compared with baseline schemes across various network configurations and numbers of coordinating adversaries. We also provide theoretical support for why eigenvector centrality-based attacks are suboptimal in decentralized FL. Overall, our findings provide valuable insights into the vulnerabilities of decentralized FL systems, setting the stage for future research aimed at developing more secure and robust decentralized FL frameworks.

cs.CR↗

arXiv2025

A Comparative Analysis of Distributed Linear Solvers under Data Heterogeneity

We consider the problem of solving a large-scale system of linear equations in a distributed or federated manner by a taskmaster and a set of machines, each possessing a subset of the equations. We provide a comprehensive comparison of two well-known classes of algorithms used to solve this problem: projection-based methods and optimization-based methods. First, we introduce a novel geometric notion of data heterogeneity called angular heterogeneity and discuss its generality. Using this notion, we characterize the optimal convergence rates of the most prominent algorithms from each class, capturing the effects of the number of machines, the number of equations, and that of both cross-machine and local data heterogeneity on these rates. Our analysis establishes the superiority of Accelerated Projected Consensus in realistic scenarios with significant data heterogeneity and offers several insights into how angular heterogeneity affects the efficiency of the methods studied. Additionally, we develop distributed algorithms for the efficient computation of the proposed angular heterogeneity metrics. Our extensive numerical analyses validate and complement our theoretical results.

cs.DC↗

arXiv2025

Decentralized Sporadic Federated Learning: A Unified Algorithmic Framework with Convergence Guarantees

Decentralized federated learning (DFL) captures FL settings where both (i) model updates and (ii) model aggregations are exclusively carried out by the clients without a central server. Existing DFL works have mostly focused on settings where clients conduct a fixed number of local updates between local model exchanges, overlooking heterogeneity and dynamics in communication and computation capabilities. In this work, we propose Decentralized Sporadic Federated Learning ($\texttt{DSpodFL}$), a DFL methodology built on a generalized notion of $\textit{sporadicity}$ in both local gradient and aggregation processes. $\texttt{DSpodFL}$ subsumes many existing decentralized optimization methods under a unified algorithmic framework by modeling the per-iteration (i) occurrence of gradient descent at each client and (ii) exchange of models between client pairs as arbitrary indicator random variables, thus capturing $\textit{heterogeneous and time-varying}$ computation/communication scenarios. We analytically characterize the convergence behavior of $\texttt{DSpodFL}$ for both convex and non-convex models and for both constant and diminishing learning rates, under mild assumptions on the communication graph connectivity, data heterogeneity across clients, and gradient noises. We show how our bounds recover existing results from decentralized gradient descent as special cases. Experiments demonstrate that $\texttt{DSpodFL}$ consistently achieves improved training speeds compared with baselines under various system settings.

cs.LG↗

arXiv2024

Optimal Interventions in Coupled-Activity Network Games: Application to Sustainable Forestry

We address the challenge of promoting sustainable practices in production forests managed by strategic entities (agents) that harvest agricultural commodities under concession agreements. These entities engage in activities that either follow sustainable production practices or expand into protected forests for agricultural growth, which leads to unsustainable production. Our study uses a network game model to design optimal pricing policies that incentivize sustainability and discourage environmentally harmful practices. Specifically, we model interactions between agents, capturing both intra-activity (within a single production activity) and cross-activity (between sustainable and unsustainable practices) influences on agent behavior. We solve the problem of maximizing welfare while adhering to budgetary and environmental constraints - particularly, limiting the aggregate level of unsustainable effort across all agents. Although this problem is NP-hard in general, we derive closed-form solutions for various realistic scenarios, including cases with regionally uniform pricing and the use of sustainability premiums or penalties. Remarkably, we find that it is possible to achieve both welfare improvement and reduction in unsustainable practices without reducing any agent's utility, even when there is no external budget for increasing premiums. We introduce a novel node centrality measure to identify agents whose decisions most influence aggregate unsustainable effort. Empirical validation confirms our theoretical findings, offering actionable insights for policymakers aiming to promote sustainable resource management in agricultural commodity markets. Our work has broader implications for addressing sustainability challenges in the presence of network effects, offering a framework for designing incentive structures that align economic objectives with environmental stewardship.

math.OC↗

arXiv2024

A Perron-Frobenius Theorem for Strongly Aperiodic Stochastic Chains

We derive a generalization of the Perron-Frobenius theorem to time-varying row-stochastic matrices as follows: using Kolmogorov's concept of absolute probability sequences, which are time-varying analogs of principal eigenvectors, we identify a set of connectivity conditions that generalize the notion of irreducibility (strong connectivity) to time-varying matrices (networks), and we show that under these conditions, the absolute probability sequence associated with a given matrix sequence is (a) uniformly positive and (b) unique. Our results apply to both discrete-time and continuous-time settings. We then discuss a few applications of our main results to non-Bayesian learning, distributed optimization, opinion dynamics, and averaging dynamics over random networks.

math.OC↗

arXiv2024

The Impact of Adversarial Node Placement in Decentralized Federated Learning Networks

As Federated Learning (FL) grows in popularity, new decentralized frameworks are becoming widespread. These frameworks leverage the benefits of decentralized environments to enable fast and energy-efficient inter-device communication. However, this growing popularity also intensifies the need for robust security measures. While existing research has explored various aspects of FL security, the role of adversarial node placement in decentralized networks remains largely unexplored. This paper addresses this gap by analyzing the performance of decentralized FL for various adversarial placement strategies when adversaries can jointly coordinate their placement within a network. We establish two baseline strategies for placing adversarial node: random placement and network centrality-based placement. Building on this foundation, we propose a novel attack algorithm that prioritizes adversarial spread over adversarial centrality by maximizing the average network distance between adversaries. We show that the new attack algorithm significantly impacts key performance metrics such as testing accuracy, outperforming the baseline frameworks by between $9\%$ and $66.5\%$ for the considered setups. Our findings provide valuable insights into the vulnerabilities of decentralized FL systems, setting the stage for future research aimed at developing more secure and robust decentralized FL frameworks.

cs.CR↗

arXiv2023

Connectivity-Aware Semi-Decentralized Federated Learning over Time-Varying D2D Networks

Semi-decentralized federated learning blends the conventional device to-server (D2S) interaction structure of federated model training with localized device-to-device (D2D) communications. We study this architecture over practical edge networks with multiple D2D clusters modeled as time-varying and directed communication graphs. Our investigation results in an algorithm that controls the fundamental trade-off between (a) the rate of convergence of the model training process towards the global optimizer, and (b) the number of D2S transmissions required for global aggregation. Specifically, in our semi-decentralized methodology, D2D consensus updates are injected into the federated averaging framework based on column-stochastic weight matrices that encapsulate the connectivity within the clusters. To arrive at our algorithm, we show how the expected optimality gap in the current global model depends on the greatest two singular values of the weighted adjacency matrices (and hence on the densities) of the D2D clusters. We then derive tight bounds on these singular values in terms of the node degrees of the D2D clusters, and we use the resulting expressions to design a threshold on the number of clients required to participate in any given global aggregation round so as to ensure a desired convergence rate. Simulations performed on real-world datasets reveal that our connectivity-aware algorithm reduces the total communication cost required to reach a target accuracy significantly compared with baselines depending on the connectivity structure and the learning task.

cs.DC↗

arXiv2022

A Random Adaptation Perspective on Distributed Averaging

We propose a random adaptation variant of time-varying distributed averaging dynamics in discrete time. We show that this leads to novel interpretations of fundamental concepts in distributed averaging, opinion dynamics, and distributed learning. Namely, we show that the ergodicity of a stochastic chain is equivalent to the almost sure (a.s.) finite-time agreement attainment in the proposed random adaptation dynamics. Using this result, we provide a new interpretation for the absolute probability sequence of an ergodic chain. We then modify the base-case dynamics into a time-reversed inhomogeneous Markov chain, and we show that in this case ergodicity is equivalent to the uniqueness of the limiting distributions of the Markov chain. Finally, we introduce and study a time-varying random adaptation version of the Friedkin-Johnsen model and a rank-one perturbation of the base-case dynamics.

math.OC↗

arXiv2022

Usefulness of the Age-Structured SIR Dynamics in Modelling COVID-19

We examine the age-structured SIR model, a variant of the classical Susceptible-Infected-Recovered (SIR) model of epidemic propagation, in the context of COVID-19. In doing so, we provide a theoretical basis for the model, perform an empirical validation, and discover the limitations of the model in approximating arbitrary epidemics. We first establish the differential equations defining the age-structured SIR model as the mean-field limits of a continuous-time Markov process that models epidemic spreading on a social network involving random, asynchronous interactions. We then show that, as the population size grows, the infection rate for any pair of age groups converges to its mean-field limit if and only if the edge update rate of the network approaches infinity, and we show how the rate of mean-field convergence depends on the edge update rate. We then propose a system identification method for parameter estimation of the bilinear ODEs of our model, and we test the model performance on a Japanese COVID-19 dataset by generating the trajectories of the age-wise numbers of infected individuals in the prefecture of Tokyo for a period of over 365 days. In the process, we also develop an algorithm to identify the different \textit{phases} of the pandemic, each phase being associated with a unique set of contact rates. Our results show a good agreement between the generated trajectories and the observed ones.

math.OC↗

arXiv2021

Non-Bayesian Social Learning on Random Digraphs with Aperiodically Varying Network Connectivity

We study non-Bayesian social learning on random directed graphs and show that under mild connectivity assumptions, all the agents almost surely learn the true state of the world asymptotically in time if the sequence of the associated weighted adjacency matrices belongs to Class $\pstar$ (a broad class of stochastic chains that subsumes uniformly strongly connected chains). We show that uniform strong connectivity, while being unnecessary for asymptotic learning, ensures that all the agents' beliefs converge to a consensus almost surely, even when the true state is not identifiable. We then provide a few corollaries of our main results, some of which apply to variants of the original update rule such as inertial non-Bayesian learning and learning via diffusion and adaptation. Others include extensions of known results on social learning. We also show that, if the network of influences is balanced in a certain sense, then asymptotic learning occurs almost surely even in the absence of uniform strong connectivity.

math.OC↗

arXiv2019

On the Convergence Properties of Social Hegselmann-Krause Dynamics

We study the convergence properties of Social Hegselmann-Krause dynamics, a {variant} of the Hegselmann-Krause (HK) model of opinion dynamics where a physical connectivity graph that accounts for the extrinsic factors that could prevent interaction between certain pairs of agents is incorporated. As opposed to the original HK dynamics (which terminate in finite time), we show that for any underlying connected and incomplete graph, under a certain mild assumption, the expected termination time of social HK dynamics is infinity. We then investigate the rate of convergence to the steady state, and provide bounds on the maximum $ε$-convergence time in terms of the properties of the physical connectivity graph. We extend this discussion and observe that for almost all $n$, there exists an $n$-vertex physical connectivity graph on which social HK dynamics may not even $ε$-converge to the steady state within a bounded time frame. We then provide nearly tight necessary and sufficient conditions for arbitrarily slow merging (a phenomenon that is essential for arbitrarily slow $ε$-convergence to the steady state). Using the necessary conditions, we show that complete $r$-partite graphs have bounded $ε$-convergence times.

math.OC↗