arXiv subjects

Ali Jadbabaie

Publications and source records attributed to Ali Jadbabaie.

At least 55 records · Page 3Linked to original sources

arXiv2023

A Non-Asymptotic Analysis of Oversmoothing in Graph Neural Networks

Oversmoothing is a central challenge of building more powerful Graph Neural Networks (GNNs). While previous works have only demonstrated that oversmoothing is inevitable when the number of graph convolutions tends to infinity, in this paper, we precisely characterize the mechanism behind the phenomenon via a non-asymptotic analysis. Specifically, we distinguish between two different effects when applying graph convolutions -- an undesirable mixing effect that homogenizes node representations in different classes, and a desirable denoising effect that homogenizes node representations in the same class. By quantifying these two effects on random graphs sampled from the Contextual Stochastic Block Model (CSBM), we show that oversmoothing happens once the mixing effect starts to dominate the denoising effect, and the number of layers required for this transition is $O(\log N/\log (\log N))$ for sufficiently dense graphs with $N$ nodes. We also extend our analysis to study the effects of Personalized PageRank (PPR), or equivalently, the effects of initial residual connections on oversmoothing. Our results suggest that while PPR mitigates oversmoothing at deeper layers, PPR-based architectures still achieve their best performance at a shallow depth and are outperformed by the graph convolution approach on certain graphs. Finally, we support our theoretical results with numerical experiments, which further suggest that the oversmoothing phenomenon observed in practice can be magnified by the difficulty of optimizing deep GNN models.

cs.LG↗

arXiv2022

Model Predictive Control via On-Policy Imitation Learning

In this paper, we leverage the rapid advances in imitation learning, a topic of intense recent focus in the Reinforcement Learning (RL) literature, to develop new sample complexity results and performance guarantees for data-driven Model Predictive Control (MPC) for constrained linear systems. In its simplest form, imitation learning is an approach that tries to learn an expert policy by querying samples from an expert. Recent approaches to data-driven MPC have used the simplest form of imitation learning known as behavior cloning to learn controllers that mimic the performance of MPC by online sampling of the trajectories of the closed-loop MPC system. Behavior cloning, however, is a method that is known to be data inefficient and suffer from distribution shifts. As an alternative, we develop a variant of the forward training algorithm which is an on-policy imitation learning method proposed by Ross et al. (2010). Our algorithm uses the structure of constrained linear MPC, and our analysis uses the properties of the explicit MPC solution to theoretically bound the number of online MPC trajectories needed to achieve optimal performance. We validate our results through simulations and show that the forward training algorithm is indeed superior to behavior cloning when applied to MPC.

math.OC↗

arXiv2022

Link Partitioning on Simplicial Complexes Using Higher-Order Laplacians

Link partitioning is a popular approach in network science used for discovering overlapping communities by identifying clusters of strongly connected links. Current link partitioning methods are specifically designed for networks modelled by graphs representing pairwise relationships. Therefore, these methods are incapable of utilizing higher-order information about group interactions in network data which is increasingly available. Simplicial complexes extend the dyadic model of graphs and can model polyadic relationships which are ubiquitous and crucial in many complex social and technological systems. In this paper, we introduce a link partitioning method that leverages higher-order (i.e. triadic and higher) information in simplicial complexes for better community detection. Our method utilizes a novel random walk on links of simplicial complexes defined by the higher-order Laplacian--a generalization of the graph Laplacian that incorporates polyadic relationships of the network. We transform this random walk into a graph-based random walk on a lifted line graph--a dual graph in which links are nodes while nodes and higher-order connections are links--and optimize for the standard notion of modularity. We show that our method is guaranteed to provide interpretable link partitioning results under mild assumptions. We also offer new theoretical results on the spectral properties of simplicial complexes by studying the spectrum of the link random walk. Experiment results on real-world community detection tasks show that our higher-order approach significantly outperforms existing graph-based link partitioning methods.

cs.SI↗

arXiv2022

Generalizing Homophily to Simplicial Complexes

Group interactions occur frequently in social settings, yet their properties beyond pairwise relationships in network models remain unexplored. In this work, we study homophily, the nearly ubiquitous phenomena wherein similar individuals are more likely than random to form connections with one another, and define it on simplicial complexes, a generalization of network models that goes beyond dyadic interactions. While some group homophily definitions have been proposed in the literature, we provide theoretical and empirical evidence that prior definitions mostly inherit properties of homophily in pairwise interactions rather than capture the homophily of group dynamics. Hence, we propose a new measure, $k$-simplicial homophily, which properly identifies homophily in group dynamics. Across 16 empirical networks, $k$-simplicial homophily provides information uncorrelated with homophily measures on pairwise interactions. Moreover, we show the empirical value of $k$-simplicial homophily in identifying when metadata on nodes is useful for predicting group interactions, whereas previous measures are uninformative.

cs.SI↗

arXiv2022

On Convergence of Gradient Descent Ascent: A Tight Local Analysis

Gradient Descent Ascent (GDA) methods are the mainstream algorithms for minimax optimization in generative adversarial networks (GANs). Convergence properties of GDA have drawn significant interest in the recent literature. Specifically, for $\min_{\mathbf{x}} \max_{\mathbf{y}} f(\mathbf{x};\mathbf{y})$ where $f$ is strongly-concave in $\mathbf{y}$ and possibly nonconvex in $\mathbf{x}$, (Lin et al., 2020) proved the convergence of GDA with a stepsize ratio $η_{\mathbf{y}}/η_{\mathbf{x}}=Θ(κ^2)$ where $η_{\mathbf{x}}$ and $η_{\mathbf{y}}$ are the stepsizes for $\mathbf{x}$ and $\mathbf{y}$ and $κ$ is the condition number for $\mathbf{y}$. While this stepsize ratio suggests a slow training of the min player, practical GAN algorithms typically adopt similar stepsizes for both variables, indicating a wide gap between theoretical and empirical results. In this paper, we aim to bridge this gap by analyzing the \emph{local convergence} of general \emph{nonconvex-nonconcave} minimax problems. We demonstrate that a stepsize ratio of $Θ(κ)$ is necessary and sufficient for local convergence of GDA to a Stackelberg Equilibrium, where $κ$ is the local condition number for $\mathbf{y}$. We prove a nearly tight convergence rate with a matching lower bound. We further extend the convergence guarantees to stochastic GDA and extra-gradient methods (EG). Finally, we conduct several numerical experiments to support our theoretical findings.

math.OC↗

arXiv2022

Beyond Worst-Case Analysis in Stochastic Approximation: Moment Estimation Improves Instance Complexity

We study oracle complexity of gradient based methods for stochastic approximation problems. Though in many settings optimal algorithms and tight lower bounds are known for such problems, these optimal algorithms do not achieve the best performance when used in practice. We address this theory-practice gap by focusing on instance-dependent complexity instead of worst case complexity. In particular, we first summarize known instance-dependent complexity results and categorize them into three levels. We identify the domination relation between different levels and propose a fourth instance-dependent bound that dominates existing ones. We then provide a sufficient condition according to which an adaptive algorithm with moment estimation can achieve the proposed bound without knowledge of noise levels. Our proposed algorithm and its analysis provide a theoretical justification for the success of moment estimation as it achieves improved instance complexity.

math.OC↗

arXiv2022

Neural Network Weights Do Not Converge to Stationary Points: An Invariant Measure Perspective

This work examines the deep disconnect between existing theoretical analyses of gradient-based algorithms and the practice of training deep neural networks. Specifically, we provide numerical evidence that in large-scale neural network training (e.g., ImageNet + ResNet101, and WT103 + TransformerXL models), the neural network's weights do not converge to stationary points where the gradient of the loss is zero. Remarkably, however, we observe that even though the weights do not converge to stationary points, the progress in minimizing the loss function halts and training loss stabilizes. Inspired by this observation, we propose a new perspective based on ergodic theory of dynamical systems to explain it. Rather than studying the evolution of weights, we study the evolution of the distribution of weights. We prove convergence of the distribution of weights to an approximate invariant measure, thereby explaining how the training loss can stabilize without weights necessarily converging to stationary points. We further discuss how this perspective can better align optimization theory with empirical observations in machine learning practice.

cs.LG↗

arXiv2022

Gradient Descent for Low-Rank Functions

Several recent empirical studies demonstrate that important machine learning tasks, e.g., training deep neural networks, exhibit low-rank structure, where the loss function varies significantly in only a few directions of the input space. In this paper, we leverage such low-rank structure to reduce the high computational cost of canonical gradient-based methods such as gradient descent (GD). Our proposed \emph{Low-Rank Gradient Descent} (LRGD) algorithm finds an $ε$-approximate stationary point of a $p$-dimensional function by first identifying $r \leq p$ significant directions, and then estimating the true $p$-dimensional gradient at every iteration by computing directional derivatives only along those $r$ directions. We establish that the "directional oracle complexities" of LRGD for strongly convex and non-convex objective functions are $\mathcal{O}(r \log(1/ε) + rp)$ and $\mathcal{O}(r/ε^2 + rp)$, respectively. When $r \ll p$, these complexities are smaller than the known complexities of $\mathcal{O}(p \log(1/ε))$ and $\mathcal{O}(p/ε^2)$ of {\gd} in the strongly convex and non-convex settings, respectively. Thus, LRGD significantly reduces the computational cost of gradient-based methods for sufficiently low-rank functions. In the course of our analysis, we also formally define and characterize the classes of exact and approximately low-rank functions.

cs.LG↗

arXiv2022

An Optimal Transport Approach to Personalized Federated Learning

Federated learning is a distributed machine learning paradigm, which aims to train a model using the local data of many distributed clients. A key challenge in federated learning is that the data samples across the clients may not be identically distributed. To address this challenge, personalized federated learning with the goal of tailoring the learned model to the data distribution of every individual client has been proposed. In this paper, we focus on this problem and propose a novel personalized Federated Learning scheme based on Optimal Transport (FedOT) as a learning algorithm that learns the optimal transport maps for transferring data points to a common distribution as well as the prediction model under the applied transport map. To formulate the FedOT problem, we extend the standard optimal transport task between two probability distributions to multi-marginal optimal transport problems with the goal of transporting samples from multiple distributions to a common probability domain. We then leverage the results on multi-marginal optimal transport problems to formulate FedOT as a min-max optimization problem and analyze its generalization and optimization properties. We discuss the results of several numerical experiments to evaluate the performance of FedOT under heterogeneous data distributions in federated learning problems.

cs.LG↗

arXiv2022

Inference in Opinion Dynamics under Social Pressure

We introduce a new opinion dynamics model where a group of agents holds two kinds of opinions: inherent and declared. Each agent's inherent opinion is fixed and unobservable by the other agents. At each time step, agents broadcast their declared opinions on a social network, which are governed by the agents' inherent opinions and social pressure. In particular, we assume that agents may declare opinions that are not aligned with their inherent opinions to conform with their neighbors. This raises the natural question: Can we estimate the agents' inherent opinions from observations of declared opinions? For example, agents' inherent opinions may represent their true political alliances (Democrat or Republican), while their declared opinions may model the political inclinations of tweets on social media. In this context, we may seek to predict the election results by observing voters' tweets, which do not necessarily reflect their political support due to social pressure. We analyze this question in the special case where the underlying social network is a complete graph. We prove that, as long as the population does not include large majorities, estimation of aggregate and individual inherent opinions is possible. On the other hand, large majorities force minorities to lie over time, which makes asymptotic estimation impossible.

eess.SY↗

arXiv2022

Byzantine-Robust Federated Linear Bandits

In this paper, we study a linear bandit optimization problem in a federated setting where a large collection of distributed agents collaboratively learn a common linear bandit model. Standard federated learning algorithms applied to this setting are vulnerable to Byzantine attacks on even a small fraction of agents. We propose a novel algorithm with a robust aggregation oracle that utilizes the geometric median. We prove that our proposed algorithm is robust to Byzantine attacks on fewer than half of agents and achieves a sublinear $\tilde{\mathcal{O}}({T^{3/4}})$ regret with $\mathcal{O}(\sqrt{T})$ steps of communication in $T$ steps. Moreover, we make our algorithm differentially private via a tree-based mechanism. Finally, if the level of corruption is known to be small, we show that using the geometric median of mean oracle for robust aggregation further improves the regret bound.

cs.LG↗

arXiv2022

Current Implicit Policies May Not Eradicate COVID-19

Successful predictive modeling of epidemics requires an understanding of the implicit feedback control strategies which are implemented by populations to modulate the spread of contagion. While this task of capturing endogenous behavior can be achieved through intricate modeling assumptions, we find that a population's reaction to case counts can be described through a second order affine dynamical system with linear control which fits well to the data across different regions and times throughout the COVID-19 pandemic. The model fits the data well both in and out of sample across the 50 states of the United States, with comparable $R^2$ scores to state of the art ensemble predictions. In contrast to recent models of epidemics, rather than assuming that individuals directly control the contact rate which governs the spread of disease, we assume that individuals control the rate at which they vary their number of interactions, i.e. they control the derivative of the contact rate. We propose an implicit feedback law for this control input and verify that it correlates with policies taken throughout the pandemic. A key takeaway of the dynamical model is that the "stable" point of case counts is non-zero, i.e. COVID-19 will not be eradicated under the current collection of policies and strategies, and additional policies are needed to fully eradicate it quickly. Hence, we suggest alternative implicit policies which focus on making interventions (such as vaccinations and mobility restrictions) a function of cumulative case counts, for which our results suggest a better possibility of eradicating COVID-19.

q-bio.PE↗

arXiv2022

Unifying Epidemic Models with Mixtures

The COVID-19 pandemic has emphasized the need for a robust understanding of epidemic models. Current models of epidemics are classified as either mechanistic or non-mechanistic: mechanistic models make explicit assumptions on the dynamics of disease, whereas non-mechanistic models make assumptions on the form of observed time series. Here, we introduce a simple mixture-based model which bridges the two approaches while retaining benefits of both. The model represents time series of cases and fatalities as a mixture of Gaussian curves, providing a flexible function class to learn from data compared to traditional mechanistic models. Although the model is non-mechanistic, we show that it arises as the natural outcome of a stochastic process based on a networked SIR framework. This allows learned parameters to take on a more meaningful interpretation compared to similar non-mechanistic models, and we validate the interpretations using auxiliary mobility data collected during the COVID-19 pandemic. We provide a simple learning algorithm to identify model parameters and establish theoretical results which show the model can be efficiently learned from data. Empirically, we find the model to have low prediction error. The model is available live at covidpredictions.mit.edu. Ultimately, this allows us to systematically understand the impacts of interventions on COVID-19, which is critical in developing data-driven solutions to controlling epidemics.

stat.ML↗

arXiv2021

Network Group Testing

We consider the problem of identifying infected individuals in a population of size N. We introduce a group testing approach that uses significantly fewer than N tests when infection prevalence is low. The most common approach to group testing, Dorfman testing, groups individuals randomly. However, as communicable diseases spread from individual to individual through underlying social networks, our approach utilizes network information to improve performance. Network grouping, which groups individuals by community, weakly dominates Dorfman testing in terms of the expected number of tests used. Network grouping's outperformance is determined by the strength of community structure in the network. When networks have strong community structure, network grouping achieves the lower bound for two-stage testing procedures. As an empirical example, we consider the scenario of a university testing its population for COVID-19. Using social network data from a Danish university, we demonstrate network grouping requires significantly fewer tests than Dorfman. In contrast to many proposed group testing approaches, network grouping is simple for practitioners to implement. In practice, individuals can be grouped by family unit, social group, or work group.

stat.AP↗

arXiv2021

Time varying regression with hidden linear dynamics

We revisit a model for time-varying linear regression that assumes the unknown parameters evolve according to a linear dynamical system. Counterintuitively, we show that when the underlying dynamics are stable the parameters of this model can be estimated from data by combining just two ordinary least squares estimates. We offer a finite sample guarantee on the estimation error of our method and discuss certain advantages it has over Expectation-Maximization (EM), which is the main approach proposed by prior work.

math.ST↗

arXiv2021

Sales-Based Rebate Design

We propose a novel family of sales-based rebate mechanisms that induce network effects in sales of products that do not exhibit such externalities. The proposed rebate mechanisms enable the seller of a product with uncertain quality to adjust the magnitude and sign of externalities in consumers' payoffs by conditioning the amount of the rebate on the sales volume. Using the machinery of global games and variational optimization techniques, we analyze the revenue implications of such induced externalities in the form of rebates. We identify optimal profitable designs while unraveling the main drivers of profit and further elucidate the main practical barriers associated with their implementation and show how these difficulties can be handled. The key insight of our rebate design is monetizing the strategic uncertainty in consumers' beliefs on others' valuations. The common externality induced in consumers' utilities as a sales-based rebate essentially enables the seller to elicit different prices at different valuations, given the heterogeneity of the beliefs on sales volume and hence on the rebate. Our analysis indicates that a mechanism that creates positive externalities will in fact reduce the profit because it lowers the expected prices at high valuations. On the other hand, sellers can use a sales-based rebate mechanism that is decreasing with sales volume to incentivize purchases at lower valuations by providing higher expected rebates. Our work contributes to the literature on technology-enabled features of digitized markets and demonstrates that real-time sales/subscription data can lead to new revenue management methods.

math.OC↗

arXiv2021

Complexity Lower Bounds for Nonconvex-Strongly-Concave Min-Max Optimization

We provide a first-order oracle complexity lower bound for finding stationary points of min-max optimization problems where the objective function is smooth, nonconvex in the minimization variable, and strongly concave in the maximization variable. We establish a lower bound of $Ω\left(\sqrtκε^{-2}\right)$ for deterministic oracles, where $ε$ defines the level of approximate stationarity and $κ$ is the condition number. Our analysis shows that the upper bound achieved in (Lin et al., 2020b) is optimal in the $ε$ and $κ$ dependence up to logarithmic factors. For stochastic oracles, we provide a lower bound of $Ω\left(\sqrtκε^{-2} + κ^{1/3}ε^{-4}\right)$. It suggests that there is a significant gap between the upper bound $\mathcal{O}(κ^3 ε^{-4})$ in (Lin et al., 2020a) and our lower bound in the condition number dependence.

math.OC↗

arXiv2021

Can Single-Shuffle SGD be Better than Reshuffling SGD and GD?

We propose matrix norm inequalities that extend the Recht-Ré (2012) conjecture on a noncommutative AM-GM inequality by supplementing it with another inequality that accounts for single-shuffle, which is a widely used without-replacement sampling scheme that shuffles only once in the beginning and is overlooked in the Recht-Ré conjecture. Instead of general positive semidefinite matrices, we restrict our attention to positive definite matrices with small enough condition numbers, which are more relevant to matrices that arise in the analysis of SGD. For such matrices, we conjecture that the means of matrix products corresponding to with- and without-replacement variants of SGD satisfy a series of spectral norm inequalities that can be summarized as: "single-shuffle SGD converges faster than random-reshuffle SGD, which is in turn faster than with-replacement SGD." We present theorems that support our conjecture by proving several special cases.

cs.LG↗