arXiv subjects

Daniel R. Figueiredo

Publications and source records attributed to Daniel R. Figueiredo.

14 recordsLinked to original sources

arXiv2023

Exact Recovery and Bregman Hard Clustering of Node-Attributed Stochastic Block Model

Network clustering tackles the problem of identifying sets of nodes (communities) that have similar connection patterns. However, in many scenarios, nodes also have attributes that are correlated with the clustering structure. Thus, network information (edges) and node information (attributes) can be jointly leveraged to design high-performance clustering algorithms. Under a general model for the network and node attributes, this work establishes an information-theoretic criterion for the exact recovery of community labels and characterizes a phase transition determined by the Chernoff-Hellinger divergence of the model. The criterion shows how network and attribute information can be exchanged in order to have exact recovery (e.g., more reliable network information requires less reliable attribute information). This work also presents an iterative clustering algorithm that maximizes the joint likelihood, assuming that the probability distribution of network interactions and node attributes belong to exponential families. This covers a broad range of possible interactions (e.g., edges with weights) and attributes (e.g., non-Gaussian models), as well as sparse networks, while also exploring the connection between exponential families and Bregman divergences. Extensive numerical experiments using synthetic data indicate that the proposed algorithm outperforms classic algorithms that leverage only network or only attribute information as well as state-of-the-art algorithms that also leverage both sources of information. The contributions of this work provide insights into the fundamental limits and practical techniques for inferring community labels on node-attributed networks.

cs.SI↗

arXiv2023

Quantum Walks can Unitarily Represent Random Walks on Finite Graphs

Quantum and random walks have been shown to be equivalent in the following sense: a time-dependent random walk can be constructed such that its vertex distribution at all time instants is identical to the vertex distribution of any discrete-time coined quantum walk on a finite graph. This equivalence establishes a deep connection between the two processes, far stronger than simply considering quantum walks as quantum analogues of classical random walks. The present work strengthens this connection by providing a construction that establishes this equivalence in the reverse direction: a unitary time-dependent quantum walk can be constructed such that its vertex distribution is identical to the vertex distribution of any random walk on a finite graph at all time instants. The construction shown here describes a quantum walk that matches a random walk without measurements at all time steps (an otherwise trivial statement): measurement is performed in a quantum walk that evolved unitarily until a given time $t$ such that its vertex distribution is identical to the random walk at time $t$. The construction procedure is general, covering both homogeneous and non-homogeneous random walks. For homogeneous random walks, unitary evolution implies time dependency for the quantum walk, since homogeneous quantum walks do not converge under arbitrary initial conditions, while a broad class of random walks does. Thus, the absence of convergence demonstrated for quantum walks in its debut comes from both time-homogeneity and unitarity, rather than unitarity alone, and our results shed light on the power of quantum walks to generate samples for arbitrary probability distributions. Finally, the construction here proposed is used to simulate quantum walks that match uniform random walks on the cycle and the torus.

quant-ph↗

arXiv2021

Heterogeneous download times in bandwidth-homogeneous BitTorrent swarms

Modeling and understanding BitTorrent (BT) dynamics is a recurrent research topic mainly due to its high complexity and tremendous practical efficiency. Over the years, different models have uncovered various phenomena exhibited by the system, many of which have direct impact on its performance. In this paper we identify and characterize a phenomenon that has not been previously observed: homogeneous peers (with respect to their upload capacities) experience heterogeneous download times. This behavior has direct impact on peer and system performance, such as high variability of download times, unfairness with respect to peer arrival order, bursty departures and content synchronization. Detailed packet-level simulations and prototype-based experiments on the Internet were performed to characterize this phenomenon. We also develop a mathematical model that accurately predicts the heterogeneous download rates of the homogeneous peers as a function of their content. In addition, we apply the model to calculate lower and upper bounds to the number of departures that occur in a burst. The heterogeneous download rates are more prevalent in unpopular swarms (very few peers). Although few works have addressed this kind of swarm, these by far represent the most common type of swarm in BT.

cs.NI↗

arXiv2020

On the equivalence between quantum and random walks on finite graphs

Quantum walks on graphs are ubiquitous in quantum computing finding a myriad of applications. Likewise, random walks on graphs are a fundamental building block for a large number of algorithms with diverse applications. While the relationship between quantum and random walks has been recently discussed in specific scenarios, this work establishes a formal equivalence between the processes on arbitrary finite graphs and general conditions for shift and coin operators. It requires empowering random walks with time heterogeneity, where the transition probability of the walker is non-uniform and time dependent. The equivalence is obtained by equating the probability of measuring the quantum walk on a given node of the graph and the probability that the random walk is at that same node, for all nodes and time steps. The result is given by the construction procedure of a matrix sequence for the random walk that yields the exact same vertex probability distribution sequence of any given quantum walk, including the scenario with multiple interfering walkers. Interestingly, these matrices allows for a different simulation approach for quantum walks where node samples respect neighbor locality and convergence is guaranteed by the law of large numbers, enabling efficient (polynomial) sampling of quantum graph trajectories (paths). Furthermore, the complexity of constructing this sequence of matrices is discussed in the general case.

quant-ph↗

arXiv2019

Building your path to escape from home

Random walks on dynamic graphs have received increasingly more attention from different academic communities over the last decade. Despite the relatively large literature, little is known about random walks that construct the graph where they walk while moving around. In this paper we study one of the simplest conceivable discrete time models of this kind, which works as follows: before every walker step, with probability $p$ a new leaf is added to the vertex currently occupied by the walker. The model grows trees and we call it the Bernoulli Growth Random Walk (BGRW). We show that the BGRW walker is transient and has a well-defined linear speed $c(p)>0$ for any $0<p\leq 1$. We also show that the tree as seen by the walker converges (in a suitable sense) to a random tree that is one-ended. Some natural open problems about this tree and variants of our model are collected at the end of the paper.

math.PR↗

arXiv2017

Transient and Slim versus Recurrent and Fat: Random Walks and the Trees they Grow

Network growth models that embody principles such as preferential attachment and local attachment rules have received much attention over the last decade. Among various approaches, random walks have been leveraged to capture such principles. In this paper we consider the No Restart Random Walk (NRRW) model where a walker builds its graph (tree) while moving around. In particular, the walker takes s steps (a parameter) on the current graph. A new node with degree one is added to the graph and connected to the node currently occupied by the walker. The walker then resumes, taking another s steps, and the process repeats. We analyze this process from the perspective of the walker and the network, showing a fundamental dichotomy between transience and recurrence for the walker as well as power law and exponential degree distribution for the network. More precisely, we prove the following results: i) for s=1 the random walk is transient and the degree of every node is bounded from above by a geometric distribution. ii) for s even the random walk is recurrent and the degree of non-leaf nodes is bounded from below by a power law distribution with exponent decreasing in s. We also provide a lower bound for the fraction of leaves in the graph, and for s=2 our bound implies that the fraction of leaves goes to one as the graph size goes to infinity. NRRW exhibits an interesting mutual dependency between graph building and random walking that is fundamentally influenced by the parity of s. Understanding this kind of coupled dynamics is an important step towards modeling more realistic network growth processes.

math.PR↗

arXiv2017

struc2vec: Learning Node Representations from Structural Identity

Structural identity is a concept of symmetry in which network nodes are identified according to the network structure and their relationship to other nodes. Structural identity has been studied in theory and practice over the past decades, but only recently has it been addressed with representational learning techniques. This work presents struc2vec, a novel and flexible framework for learning latent representations for the structural identity of nodes. struc2vec uses a hierarchy to measure node similarity at different scales, and constructs a multilayer graph to encode structural similarities and generate structural context for nodes. Numerical experiments indicate that state-of-the-art techniques for learning node representations fail in capturing stronger notions of structural identity, while struc2vec exhibits much superior performance in this task, as it overcomes limitations of prior approaches. As a consequence, numerical experiments indicate that struc2vec improves performance on classification tasks that depend more on structural identity.

cs.SI↗

arXiv2017

On the Duration and Intensity of Competitions in Nonlinear Pólya Urn Processes with Fitness

Cumulative advantage (CA) refers to the notion that accumulated resources foster the accumulation of further resources in competitions, a phenomenon that has been empirically observed in various contexts. The oldest and arguably simplest mathematical model that embodies this general principle is the Pólya urn process, which finds applications in a myriad of problems. The original model captures the dynamics of competitions between two equally fit agents under linear CA effects, which can be readily generalized to incorporate different fitnesses and nonlinear CA effects. We study two statistics of competitions under the generalized model, namely duration (i.e., time of the last tie) and intensity (i.e., number of ties). We give rigorous mathematical characterizations of the tail distributions of both duration and intensity under the various regimes for fitness and nonlinearity, which reveal very interesting behaviors. For example, fitness superiority induces much shorter competitions in the sublinear regime while much longer competitions in the superlinear regime. Our findings can shed light on the application of Pólya urn processes in more general contexts where fitness and nonlinearity may be present.

cs.PF↗

arXiv2017

Power-law decay of the degree-sequence probabilities of multiple random graphs with application to graph isomorphism

We consider events over the probability space generated by the degree sequences of multiple independent Erdős-Rényi random graphs, and consider an approximation probability space where such degree sequences are deemed to be sequences of i.i.d. random variables. We show that, for any sequence of events with probabilities asymptotically smaller than some power law in the approximation model, the same upper bound also holds in the original model. We accomplish this by extending an approximation framework proposed in a seminal paper by McKay and Wormald. Finally, as an example, we apply the developed framework to bound the probability of isomorphism-related events over multiple independent random graphs.

math.PR↗

arXiv2016

Learning Identity Mappings with Residual Gates

We propose a new layer design by adding a linear gating mechanism to shortcut connections. By using a scalar parameter to control each gate, we provide a way to learn identity mappings by optimizing only one parameter. We build upon the motivation behind Residual Networks, where a layer is reformulated in order to make learning identity mappings less problematic to the optimizer. The augmentation introduces only one extra parameter per layer, and provides easier optimization by making degeneration into identity mappings simpler. We propose a new model, the Gated Residual Network, which is the result when augmenting Residual Networks. Experimental results show that augmenting layers provides better optimization, increased performance, and more layer independence. We evaluate our method on MNIST using fully-connected networks, showing empirical indications that our augmentation facilitates the optimization of deep models, and that it provides high tolerance to full layer removal: the model retains over 90% of its performance even after half of its layers have been randomly removed. We also evaluate our model on CIFAR-10 and CIFAR-100 using Wide Gated ResNets, achieving 3.65% and 18.27% error, respectively.

cs.CV↗

arXiv2016

On the Emergence of Shortest Paths by Reinforced Random Walks

The co-evolution between network structure and functional performance is a fundamental and challenging problem whose complexity emerges from the intrinsic interdependent nature of structure and function. Within this context, we investigate the interplay between the efficiency of network navigation (i.e., path lengths) and network structure (i.e., edge weights). We propose a simple and tractable model based on iterative biased random walks where edge weights increase over time as function of the traversed path length. Under mild assumptions, we prove that biased random walks will eventually only traverse shortest paths in their journey towards the destination. We further characterize the transient regime proving that the probability to traverse non-shortest paths decays according to a power-law. We also highlight various properties in this dynamic, such as the trade-off between exploration and convergence, and preservation of initial network plasticity. We believe the proposed model and results can be of interest to various domains where biased random walks and decentralized navigation have been applied.

cs.NE↗

arXiv2016

Local symmetry in random graphs

Quite often real-world networks can be thought of as being symmetric, in the abstract sense that vertices can be found to have similar or equivalent structural roles. However, traditional measures of symmetry in graphs are based on their automorphism groups, which do not account for the similarity of local structures. We introduce the concept of local symmetry, which reflects the structural equivalence of the vertices' egonets. We study the emergence of asymmetry in the Erdős-Rényi random graph model and identify regimes of both asymptotic local symmetry and asymptotic local asymmetry. We find that local symmetry persists at least to an average degree of $n^{1/3}$ and local asymmetry emerges at an average degree not greater than $n^{1/2}$, which are regimes of much larger average degree than for traditional, global asymmetry.

math.PR↗

arXiv2014

On the duration and intensity of cumulative advantage competitions

The role of skill (fitness) and luck (randomness) as driving forces on the dynamics of resource accumulation in a myriad of systems have long puzzled scientists. Fueled by undisputed inequalities that emerge from actual competitions, there is a pressing need for better understanding the effects of skill and luck in resource accumulation. When such competitions are driven by externalities such as cumulative advantage (CA), the rich-get-richer effect, little is known with respect to fundamental properties such as their duration and intensity. In this work we provide a mathematical understanding of how CA exacerbates the role of luck in detriment of skill in simple and well-studied competition models. We show, for instance, that if two agents are competing for resources that arrive sequentially at each time unit, an early stroke of luck can place the less skilled in the lead for an extremely long period of time, a phenomenon we call "struggle of the fittest". In the absence of CA, the more skilled quickly prevails despite any early stroke of luck that the less skilled may have. We prove that duration of a simple skill and luck competition model exhibit power law tails when CA is present, regardless of skill difference, which is in sharp contrast to exponential tails when CA is absent. Our findings have important implications to competitions not only in complex social systems but also in contexts that leverage such models.

physics.soc-ph↗

arXiv2011

Heterogeneous download times in a homogeneous BitTorrent swarm

Modeling and understanding BitTorrent (BT) dynamics is a recurrent research topic mainly due to its high complexity and tremendous practical efficiency. Over the years, different models have uncovered various phenomena exhibited by the system, many of which have direct impact on its performance. In this paper we identify and characterize a phenomenon that has not been previously observed: homogeneous peers (with respect to their upload capacities) experience heterogeneous download rates. The consequences of this phenomenon have direct impact on peer and system performance, such as high variability of download times, unfairness with respect to peer arrival order, bursty departures and content synchronization. Detailed packet-level simulations and prototype-based experiments on the Internet were performed to characterize this phenomenon. We also develop a mathematical model that accurately predicts the heterogeneous download rates of the homogeneous peers as a function of their content. Although this phenomenon is more prevalent in unpopular swarms (very few peers), these by far represent the most common type of swarm in BT.

cs.NI↗