arXiv subjects

Vishwaraj Doshi

Publications and source records attributed to Vishwaraj Doshi.

11 recordsLinked to original sources

arXiv2024

Keeping Up With the Winner! Targeted Advertisement to Communities in Social Networks

When a new product enters a market already dominated by an existing product, will it survive along with this dominant product? Most of the existing works have shown the coexistence of two competing products spreading/being adopted on overlaid graphs with same set of users. However, when it comes to the survival of a weaker product on the same graph, it has been established that the stronger one dominates the market and wipes out the other. This paper makes a step towards narrowing this gap so that a new/weaker product can also survive along with its competitor with a positive market share. Specifically, we identify a locally optimal set of users to induce a community that is targeted with advertisement by the product launching company under a given budget constraint. To this end, we model the system as competing Susceptible-Infected-Susceptible (SIS) epidemics and employ perturbation techniques to quantify and attain a positive market share in a cost-efficient manner. Our extensive simulation results with real-world graph dataset show that with our choice of target users, a new product can establish itself with positive market share, which otherwise would be dominated and eventually wiped out of the competitive market under the same budget constraint.

eess.SY↗

arXiv2024

Accelerating Distributed Stochastic Optimization via Self-Repellent Random Walks

We study a family of distributed stochastic optimization algorithms where gradients are sampled by a token traversing a network of agents in random-walk fashion. Typically, these random-walks are chosen to be Markov chains that asymptotically sample from a desired target distribution, and play a critical role in the convergence of the optimization iterates. In this paper, we take a novel approach by replacing the standard linear Markovian token by one which follows a nonlinear Markov chain - namely the Self-Repellent Radom Walk (SRRW). Defined for any given 'base' Markov chain, the SRRW, parameterized by a positive scalar {\alpha}, is less likely to transition to states that were highly visited in the past, thus the name. In the context of MCMC sampling on a graph, a recent breakthrough in Doshi et al. (2023) shows that the SRRW achieves O(1/{\alpha}) decrease in the asymptotic variance for sampling. We propose the use of a 'generalized' version of the SRRW to drive token algorithms for distributed stochastic optimization in the form of stochastic approximation, termed SA-SRRW. We prove that the optimization iterate errors of the resulting SA-SRRW converge to zero almost surely and prove a central limit theorem, deriving the explicit form of the resulting asymptotic covariance matrix corresponding to iterate errors. This asymptotic covariance is always smaller than that of an algorithm driven by the base Markov chain and decreases at rate O(1/{\alpha}^2) - the performance benefit of using SRRW thereby amplified in the stochastic optimization context. Empirical results support our theoretical findings.

math.PR↗

arXiv2024

Central Limit Theorem for Two-Timescale Stochastic Approximation with Markovian Noise: Theory and Applications

Two-timescale stochastic approximation (TTSA) is among the most general frameworks for iterative stochastic algorithms. This includes well-known stochastic optimization methods such as SGD variants and those designed for bilevel or minimax problems, as well as reinforcement learning like the family of gradient-based temporal difference (GTD) algorithms. In this paper, we conduct an in-depth asymptotic analysis of TTSA under controlled Markovian noise via central limit theorem (CLT), uncovering the coupled dynamics of TTSA influenced by the underlying Markov chain, which has not been addressed by previous CLT results of TTSA only with Martingale difference noise. Building upon our CLT, we expand its application horizon of efficient sampling strategies from vanilla SGD to a wider TTSA context in distributed learning, thus broadening the scope of Hu et al. (2022). In addition, we leverage our CLT result to deduce the statistical properties of GTD algorithms with nonlinear function approximation using Markovian samples and show their identical asymptotic performance, a perspective not evident from current finite-time bounds.

stat.ML↗

arXiv2023

Self-Repellent Random Walks on General Graphs -- Achieving Minimal Sampling Variance via Nonlinear Markov Chains

We consider random walks on discrete state spaces, such as general undirected graphs, where the random walkers are designed to approximate a target quantity over the network topology via sampling and neighborhood exploration in the form of Markov chain Monte Carlo (MCMC) procedures. Given any Markov chain corresponding to a target probability distribution, we design a self-repellent random walk (SRRW) which is less likely to transition to nodes that were highly visited in the past, and more likely to transition to seldom visited nodes. For a class of SRRWs parameterized by a positive real {\alpha}, we prove that the empirical distribution of the process converges almost surely to the the target (stationary) distribution of the underlying Markov chain kernel. We then provide a central limit theorem and derive the exact form of the arising asymptotic co-variance matrix, which allows us to show that the SRRW with a stronger repellence (larger {\alpha}) always achieves a smaller asymptotic covariance, in the sense of Loewner ordering of co-variance matrices. Especially for SRRW-driven MCMC algorithms, we show that the decrease in the asymptotic sampling variance is of the order O(1/{\alpha}), eventually going down to zero. Finally, we provide numerical simulations complimentary to our theoretical results, also empirically testing a version of SRRW with {\alpha} increasing in time to combine the benefits of smaller asymptotic variance due to large {\alpha}, with empirically observed faster mixing properties of SRRW with smaller {\alpha}.

math.PR↗

arXiv2022

Convergence of Bi-Virus Epidemic Models with Non-Linear Rates on Networks -- A Monotone Dynamical Systems Approach

We study convergence properties of competing epidemic models of the Susceptible-Infected-Susceptible (SIS) type. The SIS epidemic model has seen widespread popularity in modelling the spreading dynamics of contagions such as viruses, infectious diseases, or even rumors/opinions over contact networks (graphs).We analyze the case of two such viruses spreading on overlaid graphs, with non-linear rates of infection spread and recovery. We call this the non-linear bi-virus model and, building upon recent results, obtain precise conditions for global convergence of the solutions to a trichotomy of possible outcomes: a virus-free state, a single-virus state, and to a coexistence state. Our techniques are based on the theory of monotone dynamical systems (MDS), in contrast to Lyapunov based techniques that have only seen partial success in determining convergence properties in the setting of competing epidemics. We demonstrate how the existing works have been unsuccessful in characterizing a large subset of the model parameter space for bi-virus epidemics, including all scenarios leading to coexistence of the epidemics. To the best of our knowledge, our results are the first in providing complete convergence analysis for the bi-virus system with nonlinear infection and recovery rates on general graphs.

eess.SY↗

arXiv2022

Minimizing File Transfer Time in Opportunistic Spectrum Access Model

We study the file transfer problem in opportunistic spectrum access (OSA) model, which has been widely studied in throughput-oriented applications for max-throughput strategies and in delay-related works that commonly assume identical channel rates and fixed file sizes. Our work explicitly considers minimizing the file transfer time for a given file in a set of heterogeneous-rate Bernoulli channels, showing that max-throughput policy doesn't minimize file transfer time in general. We formulate a mathematical framework for static extend to dynamic policies by mapping our file transfer problem to a stochastic shortest path problem. We analyze the performance of our proposed static and dynamic optimal policies over the max-throughput policy. We propose a mixed-integer programming formulation as an efficient alternative way to obtain the dynamic optimal policy and show a huge reduction in computation time. Then, we propose a heuristic policy that takes into account the performance-complexity tradeoff and consider the online implementation with unknown channel parameters. Furthermore, we present numerical simulations to support our analytical results and discuss the effect of switching delay on different policies. Finally, we extend the file transfer problem to Markovian channels and demonstrate the impact of the correlation of each channel.

math.OC↗

arXiv2022

Bi-SIS Epidemics on Graphs -- Quantitative Analysis of Coexistence Equilibria

We consider a system in which two viruses of the Susceptible-Infected-Susceptible (SIS) type compete over general, overlaid graphs. While such systems have been the focus of many recent works, they have mostly been studied in the sense of convergence analysis, with no existing results quantifying the non-trivial coexistence equilibria (CE) - that is, when both competing viruses maintain long term presence over the network. In this paper, we prove monotonicity of the CE with respect to effective infection rates of the two viruses, and provide the first quantitative analysis of such equilibria in the form of upper bounds involving spectral radii of the underlying graphs, as well as positive equilibria of related single-virus systems. Our results provide deeper insight into how the long term infection probabilities are affected by system parameters, which we further highlight via numerical results.

eess.SY↗

arXiv2022

Efficiency Ordering of Stochastic Gradient Descent

We consider the stochastic gradient descent (SGD) algorithm driven by a general stochastic sequence, including i.i.d noise and random walk on an arbitrary graph, among others; and analyze it in the asymptotic sense. Specifically, we employ the notion of `efficiency ordering', a well-analyzed tool for comparing the performance of Markov Chain Monte Carlo (MCMC) samplers, for SGD algorithms in the form of Loewner ordering of covariance matrices associated with the scaled iterate errors in the long term. Using this ordering, we show that input sequences that are more efficient for MCMC sampling also lead to smaller covariance of the errors for SGD algorithms in the limit. This also suggests that an arbitrarily weighted MSE of SGD iterates in the limit becomes smaller when driven by more efficient chains. Our finding is of particular interest in applications such as decentralized optimization and swarm learning, where SGD is implemented in a random walk fashion on the underlying communication graph for cost issues and/or data privacy. We demonstrate how certain non-Markovian processes, for which typical mixing-time based non-asymptotic bounds are intractable, can outperform their Markovian counterparts in the sense of efficiency ordering for SGD. We show the utility of our method by applying it to gradient descent with shuffling and mini-batch gradient descent, reaffirming key results from existing literature under a unified framework. Empirically, we also observe efficiency ordering for variants of SGD such as accelerated SGD and Adam, open up the possibility of extending our notion of efficiency ordering to a broader family of stochastic optimization algorithms.

cs.LG↗

arXiv2021

Opportunistic Spectrum Access: Does Maximizing Throughput Minimize File Transfer Time?

The Opportunistic Spectrum Access (OSA) model has been developed for the secondary users (SUs) to exploit the stochastic dynamics of licensed channels for file transfer in an opportunistic manner. Common approaches to design channel sensing strategies for throughput-oriented applications tend to maximize the long-term throughput, with the hope that it provides reduced file transfer time as well. In this paper, we show that this is not correct in general, especially for small files. Unlike prior delay-related works that seldom consider the heterogeneous channel rate and bursty incoming packets, our work explicitly considers minimizing the file transfer time of a single file consisting of multiple packets in a set of heterogeneous channels. We formulate a mathematical framework for the static policy, and extend to dynamic policy by mapping our file transfer problem to the stochastic shortest path problem. We analyze the performance of our proposed static optimal and dynamic optimal policies over the policy that maximizes long-term throughput. We then propose a heuristic policy that takes into account the performance-complexity tradeoff and an extension to online implementation with unknown channel parameters, and also present the regret bound for our online algorithm. We also present numerical simulations that reflect our analytical results.

math.OC↗

arXiv2021

Competing Epidemics on Graphs -- Global Convergence and Coexistence

The dynamics of the spread of contagions such as viruses, infectious diseases or even rumors/opinions over contact networks (graphs) have effectively been captured by the well known \textit{Susceptible-Infected-Susceptible} ($SIS$) epidemic model in recent years. When it comes to competition between two such contagions spreading on overlaid graphs, their propagation is captured by so-called \textit{bi-virus} epidemic models. Analysis of such dynamical systems involve the identification of equilibrium points and its convergence properties, which determine whether either of the viruses dies out, or both survive together. We demonstrate how the existing works are unsuccessful in characterizing a large subset of the model parameter space, including all parameters for which the competitiveness of the bi-virus system is significant enough to attain coexistence of the epidemics. In this paper, we fill in this void and obtain convergence results for the entirety of the model parameter space; giving precise conditions (necessary and sufficient) under which the system \textit{globally converges} to a \textit{trichotomy} of possible outcomes: a virus-free state, a single-virus state, and to a coexistence state -- the first such result.

eess.SY↗

arXiv2020

Fiedler Vector Approximation via Interacting Random Walks

The Fiedler vector of a graph, namely the eigenvector corresponding to the second smallest eigenvalue of a graph Laplacian matrix, plays an important role in spectral graph theory with applications in problems such as graph bi-partitioning and envelope reduction. Algorithms designed to estimate this quantity usually rely on a priori knowledge of the entire graph, and employ techniques such as graph sparsification and power iterations, which have obvious shortcomings in cases where the graph is unknown, or changing dynamically. In this paper, we develop a framework in which we construct a stochastic process based on a set of interacting random walks on a graph and show that a suitably scaled version of our stochastic process converges to the Fiedler vector for a sufficiently large number of walks. Like other techniques based on exploratory random walks and on-the-fly computations, such as Markov Chain Monte Carlo (MCMC), our algorithm overcomes challenges typically faced by power iteration based approaches. But, unlike any existing random walk based method such as MCMCs where the focus is on the leading eigenvector, our framework with interacting random walks converges to the Fiedler vector (second eigenvector). We also provide numerical results to confirm our theoretical findings on different graphs, and show that our algorithm performs well over a wide range of parameters and the number of random walks. Simulations results over time varying dynamic graphs are also provided to show the efficacy of our random walk based technique in such settings. As an important contribution, we extend our results and show that our framework is applicable for approximating not just the Fiedler vector of graph Laplacians, but also the second eigenvector of any time reversible Markov Chain kernel via interacting random walks.

math.PR↗