Models predict community interest in news articles with high accuracy.
problem Predict community-specific interest in news articles using content features.
method Computed content-based features on 60K articles from 4 communities. Trained and tested models over 3 time periods.
result Models can classify articles into communities with high accuracy (ROC AUC 0.81 to 1.0).
Distributed stochastic gradient descent is an important subroutine in distributed learning. A setting of particular interest is when the clients are mobile devices, where two important concerns are communication efficiency and the privacy of the clients. Several recent works have focused on reducing the communication c…
Community detection is a central problem of network data analysis. Given a network, the goal of community detection is to partition the network nodes into a small number of clusters, which could often help reveal interesting structures. The present paper studies community detection in Degree-Corrected Block Models (DCB…
Paper introduces a hyperbolic approach for community detection on graphs.
problem Detecting communities on graphs efficiently and effectively.
method Combines hyperbolic embeddings with Riemannian K-means or mixture models.
result Demonstrates effectiveness through experiments on real-world social networks.
Study uniform learnability of binary classification networks with communication.
problem Learning a network with communication between vertices from uniform ergodic Random Graph Process.
method Introduced structural Rademacher complexity and used martingale method and Marton's coupling.
result Uniform learnability as worst-case theoretical limits for binary classification problems.
Paper improves communication in decentralized federated learning for EHRs.
problem Efficiently learn from large, decentralized EHR databases.
method Fully decentralized federated learning with iterative local updates and reduced communication rounds.
result Significant reduction in communication rounds without compromising solution optimality.
A new test optimizes detecting small communities in large networks.
problem Detecting small communities in large networks.
method Using Sinkhorn's theorem and a degree-corrected block model (DCBM), the study optimizes the SgnQ test for this challenging setting.
result The SgnQ test is optimal for detecting communities larger than √n, achieving the computational lower bound (CLB).
The last decade has seen great progress in both dynamic network modeling and topic modeling. This paper draws upon both areas to create a Bayesian method that allows topic discovery to inform the latent network model and the network structure to facilitate topic identification. We apply this method to the 467 top polit…
A hierarchical gamma process infinite edge partition model is proposed to factorize the binary adjacency matrix of an unweighted undirected relational network under a Bernoulli-Poisson link. The model describes both homophily and stochastic equivalence, and is scalable to big sparse networks by focusing its computation…
Paper studies community detection in censored hypergraphs using information theory.
problem Community detection in censored hypergraphs with missing values.
method Information-theoretic approach, polynomial-time algorithm, spectral algorithm with refinement.
result Derives information-theoretic threshold for exact recovery of community structure.
Study shows it's impossible to count communities without finding them.
problem Determining the number and sizes of communities in random graph models.
method Hypothesis testing between models with different community structures, using low-degree polynomial framework.
result Testing between two different planted distributions is as hard as finding the communities.
Recently, there has been an increasing interest in designing distributed convex optimization algorithms under the setting where the data matrix is partitioned on features. Algorithms under this setting sometimes have many advantages over those under the setting where data is partitioned on samples, especially when the …
Study on efficient estimation of Gaussian mean with limited communication.
problem Estimating Gaussian mean under communication constraints.
method Decomposition into localization and refinement stages, development of communication-efficient and statistically optimal procedures.
result Established minimax rates of convergence and developed optimal procedures.
The paper introduces curvature-based clustering algorithms for graph analysis.
problem Identifying densely connected substructures in graphs for community detection.
method Discrete Ricci curvatures and geometric flows to reveal community structure.
result The curvature-based approach can identify overlapping communities in graphs.
For many networks of scientific interest we know both the connections of the network and information about the network nodes, such as the age or gender of individuals in a social network, geographic location of nodes in the Internet, or cellular function of nodes in a gene regulatory network. Here we demonstrate how th…
In distributed systems, communication is a major concern due to issues such as its vulnerability or efficiency. In this paper, we are interested in estimating sparse inverse covariance matrices when samples are distributed into different machines. We address communication efficiency by proposing a method where, in a si…
Study examines Fed's pandemic communication strategies.
problem Analyzing Federal Reserve's communication during the COVID-19 pandemic.
method Sentiment analysis, topic modeling, comparative analysis of previous crises.
result Fed's communication during the pandemic focused on financial stability, market volatility, social welfare, and unconventional monetary policy.
New method reduces communication costs in distributed nonconvex optimization.
problem Large communication costs between central server and local workers in distributed learning.
method Communication-compressed AMSGrad for distributed nonconvex optimization.
result Converges to first-order stationary point with same iteration complexity as vanilla AMSGrad.
Federated learning enables ML in wireless without central data.
problem Centralized ML in wireless is impractical due to data privacy and overhead.
method Federated learning, where data stays local and ML is done remotely.
result Federated learning is suitable for 5G and other wireless applications.
Survey of convex optimization for community detection in networks.
problem Community detection in networks.
method Convex optimization techniques and their theoretical analysis.
result Advantages of convex community detection, including robustness, consistency, and adaptivity.
Study hypothesis testing under quantized samples with communication constraints, achieving near-optimal sample complexity.
problem Optimizing hypothesis testing with quantized samples and communication constraints.
method Developed a polynomial-time algorithm achieving near-optimal sample complexity under communication constraints.
result Achieved near-optimal sample complexity under communication constraints, with a logarithmic factor increase over unconstrained setting.
Community detection using both graphs and social networks is the focus of many algorithms. Recent methods aimed at optimizing the so-called modularity function proceed by maximizing relations within communities while minimizing inter-community relations. However, given the NP-completeness of the problem, these algorith…
Media seems to have become more partisan, often providing a biased coverage of news catering to the interest of specific groups. It is therefore essential to identify credible information content that provides an objective narrative of an event. News communities such as digg, reddit, or newstrust offer recommendations,…
Paper proposes SCALLION and SCAFCOM for compressed FL with reduced communication.
problem Reducing communication overhead in Federated Learning with data heterogeneity and partial participation.
method Revisit and simplify stochastic controlled averaging, proposing SCALLION and SCAFCOM for unbiased and biased compression.
result SCALLION and SCAFCOM outperform existing methods in communication and computation complexities.
Estimating the largest community in a mixed population via sequential sampling.
problem Identifying the largest community in a mixed population with limited sampling.
method Sequential, random sampling of individuals across multiple boxes, optimizing sampling strategy and decision rule.
result Proposed algorithms achieve optimal error probability decay rates under fixed budget constraints.
Paper proposes DMGD for integrating outlier and community detection in graph embedding.
problem Outlier nodes affect graph embedding of regular nodes, especially in networks with multiple communities.
method DMGD integrates outlier and community detection with node embedding using multiclass graph description.
result DMGD detects outliers relative to their communities and achieves better node embedding compared to state-of-the-arts.
Attention to entropic communication improves message decoding and cooperation.
problem Improper weighting in attention mechanisms hinders optimal communication.
method Combining attention and relative entropy (RE) for optimal message encoding and decoding.
result Proper attention communication emerges via weighted RE, aiding optimal protocols.
TWIST algorithm detects communities in multi-layer networks with tensor decomposition.
problem Community detection in multi-layer networks with multiple node-modality relationships.
method Tensor-based TWIST algorithm for global/local node and layer memberships.
result Accurate community detection with small misclassification error as network size increases.
The paper explores distributed learning with minimal communication in high-dimensional settings.
problem Learning in high-dimensional settings with sublinear communication.
method Combining mirror descent with randomized sparsification/quantization of iterates.
result It is possible to learn a d-dimensional model in the distributed setting with logarithmic communication in d. In standard graph clustering/community detection, one is interested in partitioning the graph into more densely connected subsets of nodes. In contrast, the "search" problem of this paper aims to only find the nodes in a "single" such community, the target, out of the many communities that may exist. To do so , we are …
Paper models graph edge dependencies using latent variables for community detection.
problem Graphs' edge dependencies not fully explained by community membership.
method Introduces auxiliary latent variables to model edge dependencies and analyzes conditions for exact recovery.
result Exact recovery possible by semidefinite programming down to maximum likelihood threshold.
New model detects communities in multiplex networks, accounting for layer dependencies.
problem Detecting communities in multiplex networks with layer-specific dependencies.
method Hierarchical Bayesian model with a hierarchical Dirichlet prior and slice sampler.
result Model automatically picks the number of communities at each layer, outperforming single-layer alternatives.
Optimal joint separation condition for radar and communications channels in dual-blind deconvolution.
problem Recovering information from overlaid radar and communications signals with unknown channels.
method Extremal functions from Beurling-Selberg interpolation theory for joint separation, nuclear norm minimization for matrix retrieval, and MUSIC for parameter estimation.
result Guaranteed well-conditioned Vandermonde matrix for MUSIC, validating theoretical findings.
Proposes a new model to capture interaction modalities in networks.
problem Limited information in binary network representations of interactions.
method Topic Blockmodel using a count-based topic model.
result Better extraction of information from interaction networks.
We study the following generalized matrix rank estimation problem: given an n×n matrix and a constant c≥0, estimate the number of eigenvalues that are greater than c. In the distributed setting, the matrix of interest is the sum of m matrices held by separate machines. We show that any deterministic…
Paper addresses privacy and communication in distributed learning, achieving optimal performance.
problem Balancing privacy, communication, and accuracy in distributed learning and estimation.
method Developed novel encoding and decoding mechanisms for mean and frequency estimation under local differential privacy and communication constraints.
result Achieved optimal privacy and communication efficiency in mean and frequency estimation.
To capture the inherent geometric features of many community detection problems, we propose to use a new random graph model of communities that we call a Geometric Block Model. The geometric block model generalizes the random geometric graphs in the same way that the well-studied stochastic block model generalizes the …
Bayesian model for sparse graphs with flexible degree distribution and overlapping communities.
problem Modeling sparse graphs with flexible degree distributions and overlapping communities.
method Non-projective inhomogeneous random graph models with specific link probability for scalable posterior inference.
result The model can provide a good fit to the degree distribution and recover well the latent community structure.
ABCD model generates synthetic graphs for community detection with improved scalability and interpretability.
problem Synthetic graphs for community detection are limited in scalability and interpretability.
method Developed a new random graph model (ABCD) with community structure and power-law distribution.
result ABCD model solves scalability and interpretability issues of LFR model.
GraphHull models networks with clear multi-scale explanations of community structure.
problem Lack of self-explainable models in graph machine learning.
method Two-level convex hulls with global archetypes and local prototypes.
result GraphHull models networks with clear multi-scale explanations.
One-shot algorithm for feature-distributed kernel PCA reduces communication costs.
problem Efficiently perform kernel PCA in distributed computing environments.
method Inspired by dual relationship between sample-distributed and feature-distributed scenarios, proposes a one-shot algorithm for feature-distributed kernel PCA.
result The algorithm provides high-quality results with low communication costs, especially when eigenvalues decay fast.
A new method reduces communication in distributed learning by 32x.
problem High communication costs in distributed machine learning.
method Distributed compressed SGD with Nesterov's momentum and blockwise compression.
result Achieves 32x reduction in communication cost and converges as fast as full-precision SGD.
Many algorithms have been proposed for fitting network models with communities, but most of them do not scale well to large networks, and often fail on sparse networks. Here we propose a new fast pseudo-likelihood method for fitting the stochastic block model for networks, as well as a variant that allows for an arbitr…
Introduces challenges and techniques for creating machine translation for indigenous languages.
problem Limited data for machine translation of indigenous languages.
method Introduction to challenges, concepts, and techniques for creating MT systems.
result Discussion of recent advances and open questions in NLP for these languages.
We study the tradeoff between the statistical error and communication cost of distributed statistical estimation problems in high dimensions. In the distributed sparse Gaussian mean estimation problem, each of the m machines receives n data points from a d-dimensional Gaussian distribution with unknown mean θ w…
Adversarial machine learning hides 5G communications from eavesdroppers.
problem Hiding 5G communications from eavesdroppers using DL classifiers.
method Craft adversarial perturbations to fool DL classifiers.
result Adversarial perturbations can hide 5G communications with high probability.
Method predicts brain regions for neuroimaging phenotypes.
problem Predicting phenotypes from brain networks without static community structure.
method Supervised community detection using block-structured regularization and ADMM optimization.
result The method identifies task-specific brain regions that improve phenotype prediction.
Spectral algorithms solve optimal community detection and related problems.
problem Optimal detection of community structures and related substructures.
method Spectral algorithms applied to various planted substructures.
result Spectral algorithms achieve optimal performance for a wide range of planted substructures.