This work presents a comprehensive understanding of the estimation of a planted low-rank signal from a general spiked tensor model near the computational threshold. Relying on standard tools from the theory of large random matrices, we characterize the large-dimensional spectral behavior of the unfoldings of the data tensor and exhibit relevant signal-to-noise ratios governing the detectability of the principal directions of the signal. These results allow to accurately predict the reconstruction performance of truncated multilinear SVD (MLSVD) in the non-trivial regime. This is particularly important since it serves as an initialization of the higher-order orthogonal iteration (HOOI) scheme, whose convergence to the best low-multilinear-rank approximation depends entirely on its initialization. We give a sufficient condition for the convergence of HOOI and show that the number of iterations before convergence tends to $1$ in the large-dimensional limit.

The performance of spectral clustering relies on the fluctuations of the entries of the eigenvectors of a similarity matrix, which has been left uncharacterized until now. In this letter, it is shown that thesignal$+$noisestructure of a general spike random matrix model is transferred to the eigenvectors of the corresponding Gram kernel matrix and the fluctuations of their entries are Gaussian in the large-dimensional regime. This CLT-like result was the last missing piece to precisely predict the classification performance of spectral clustering. The proposed proof is very general and relies solely on the rotational invariance of the noise. Numerical experiments on synthetic and real data illustrate the universality of this phenomenon.

The article considers semi-supervised multitask learning on a Gaussian mixture model (GMM). Using methods from statistical physics, we compute the asymptotic Bayes risk of each task in the regime of large datasets in high dimension, from which we analyze the role of task similarity in learning and evaluate the performance gain when tasks are learned together rather than separately. In the supervised case, we derive a simple algorithm that attains the Bayes optimal performance.

This article analyzes a kernel-based transfer learning method, under a k-class Gaussian mixture model for the input data. Following recent advances in random matrix theory, we propose new insights in transfer learning schemes for challenging cases, when the first-order statistics of all data classes coincide. The article proves the asymptotic normality of the LS-SVM decision function for any smooth kernel function. As a result, an optimization scheme is proposed to minimize the classification error rate. Our theoretical results are corroborated through simulations and then successfully applied to the context of transfer learning for PolSAR image classification.

The article proposes and theoretically analyses a computationally efficient multi-task learning (MTL) extension of popular principal component analysis (PCA)-based supervised learning schemes. The analysis reveals that (i) by default, learning may dramatically fail by suffering from negative transfer, but that (ii) simple counter-measures on data labels avert negative transfer and necessarily result in improved performances. Supporting experiments on synthetic and real data benchmarks show that the proposed method achieves comparable performance with state-of-the-art MTL methods but at a significantly reduced computational cost.

This article introduces a random matrix framework for the analysis of clustering on high-dimensional data streams, a particularly relevant setting for a more sober processing of large amounts of data with limited memory and energy resources. Assuming data $\mathbf{x}_1, \mathbf{x}_2, \ldots$ arrives as a continuous flow and a small number $L$ of them can be kept in the learning pipeline, one has only access to the diagonal elements of the Gram kernel matrix: $\left[ \mathbf{K}_L \right]_{i, j} = \frac{1}{p} \mathbf{x}_i^\top \mathbf{x}_j \mathbf{1}_{\left\lvert i - j \right\rvert < L}$. Under a large-dimensional data regime, we derive the limiting spectral distribution of the banded kernel matrix $\mathbf{K}_L$ and study its isolated eigenvalues and eigenvectors, which behave in an unfamiliar way. We detail how these results can be used to perform efficient online kernel spectral clustering and provide theoretical performance guarantees. Our findings are empirically confirmed on image clustering tasks. Leveraging on optimality results of spectral methods for clustering, this work offers insights on efficient online clustering techniques for high-dimensional data.

Several machine learning problems such as latent variable model learning and community detection can be addressed by estimating a low-rank signal from a noisy tensor. Despite recent substantial progress on the fundamental limits of the corresponding estimators in the large-dimensional setting, some of the most significant results are based on spin glass theory, which is not easily accessible to non-experts. We propose a sharply distinct and more elementary approach, relying on tools from random matrix theory. The key idea is to study random matrices arising from contractions of a random tensor, which give access to its spectral properties. In particular, for a symmetric $d$th-order rank-one model with Gaussian noise, our approach yields a novel characterization of maximum likelihood (ML) estimation performance in terms of a fixed-point equation valid in the regime where weak recovery is possible. For $d=3$, the solution to this equation matches the existing results. We conjecture that the same holds for any order $d$, based on numerical evidence for $d \in \{4,5\}$. Moreover, our analysis illuminates certain properties of the large-dimensional ML landscape. Our approach can be extended to other models, including asymmetric and non-Gaussian ones.

This paper provides a large dimensional analysis of the Softmax classifier. We discover and prove that, when the classifier is trained on data satisfying loose statistical modeling assumptions, its weights become deterministic and solely depend on the data statistical means and covariances. As a striking consequence, despite the implicit and non-linear nature of the underlying optimization problem, the performance of the Softmax classifier is the same as if performed on a mere Gaussian mixture model, thereby disrupting the intuition that non-linearities inherently extract advanced statistical features from the data. Our findings are theoretically as well as numerically sustained on CNN representations of images produced by GANs.

Stochastic Neighbour Embedding methods (SNE, t-SNE) aim at finding a faithful low-dimensional representation of a high-dimensional dataset. Despite their popularity, being solution to a non-convex optimization, the behavior of these tools is not well understood. This work provides first answers by leveraging a large dimensional statistics approach, where the number n and dimension p of the large-dimensional data are of the same magnitude. We derive and study the canonical equation verified by the critical points of this non-convex optimization problem. The study notably reveals that, in a simple setup, the achievable SNE solutions correspond to a subset of those critical points. In particular, when the clusters composing the dataset are balanced in size, these solutions are symmetrical and assume closed-form expressions.As a major conclusion, the analysis rigorously proves a long-standing heuristic statement on the "proper normalization" of the symmetric SNE: out of two natural normalization choices, only the claimed proper one leads to non-trivial solutions.

The article introduces an elementary cost and storage reduction method for spectral clustering and principal component analysis. The method consists in randomly “puncturing” both the data matrix $X\in\mathbb{C}^{p\times n}$ (or $\mathbb{R}^{p\times n}$) and its corresponding kernel (Gram) matrix $K$ through Bernoulli masks: $S\in\{0,1\}^{p\times n}$ for $X$ and $B\in\{0,1\}^{n\times n}$ for $K$. The resulting “two-way punctured” kernel is thus given by $K=\frac1p[(X\odot S)^\H (X\odot S)]\odot B$. We demonstrate that, for $X$ composed of independent columns drawn from a Gaussian mixture model, as $n,p\to\infty$ with $p/n\to c_0\in(0,\infty)$, the spectral behavior of $K$ – its limiting eigenvalue distribution, as well as its isolated eigenvalues and eigenvectors – is fully tractable and exhibits a series of counter-intuitive phenomena. We notably prove, and empirically confirm on various image databases, that it is possible to drastically puncture the data, thereby providing possibly huge computational and storage gains, for a virtually constant (clustering or PCA) performance. This preliminary study opens as such the path towards rethinking, from a large dimensional standpoint, computational and storage costs in elementary machine learning models.

This article considers spectral community detection in the regime of sparse networks with heterogeneous degree distributions, for which we devise an algorithm to efficiently retrieve communities. Specifically, we demonstrate that a well parametrized form of regularized Laplacian matrices can be used to perform spectral clustering in sparse networks without suffering from its degree heterogeneity. Besides, we exhibit important connections between this proposed matrix and the now popular non-backtracking matrix, the Bethe-Hessian matrix, as well as the standard Laplacian matrix. Interestingly, as opposed to competitive methods, our proposed improved parametrization inherently accounts for the hardness of the classification problem. These findings are summarized under the form of an algorithm capable of both estimating the number of communities and achieving high-quality community reconstruction.

Semi-supervised Laplacian regularization, a standard graph-based approach for learning from both labelled and unlabelled data, was recently demonstrated to have an insignificant high dimensional learning efficiency with respect to unlabelled data, causing it to be outperformed by its unsupervised counterpart, spectral clustering, given sufficient unlabelled data. Following a detailed discussion on the origin of this inconsistency problem, a novel regularization approach involving centering operation is proposed as solution, supported by both theoretical analysis and empirical results.

This article establishes that, unlike the legacy tf*idf representation, recent natural language representations (word embedding vectors) tend to exhibit a so-called concentration of measure phenomenon, in the sense that, as the representation size p and database size n are both large, their behavior is similar to that of large dimensional Gaussian random vectors.This phenomenon may have important consequences as machine learning algorithms for natural language data could be amenable to improvement, thereby providing new theoretical insights into the field of natural language processing.

Regularization of the classical Laplacian matrices was empirically shown to improve spectral clustering in sparse networks. It was observed that small regularizations are preferable, but this point was left as a heuristic argument. In this paper we formally determine a proper regularization which is intimately related to alternative state-of-the-art spectral techniques for sparse graphs.

Inspired by human learning, which transfers knowledge from learned tasks to solve new tasks, multitask learning aims at simultaneously solving multiple tasks by a smart exploitation of their similarities. How to relate the tasks so to optimize their performances is however a largely open problem.Based on a random matrix approach, this article proposes an asymptotic analysis of a support vector machine-inspired multitask learning scheme. The asymptotic performance of the algorithm, validated on both synthetic and real data, sets forth the relation between the statistics of the data in each task and the hyperparameters relating the tasks together. The article, as such, provides first insights on an offline control of multitask learning, which finds natural connections to the currently popular transfer learning paradigm.

This paper shows that deep learning (DL) representations of data produced by generative adversarial nets (GANs) are random vectors which fall within the class of so-called \emph{concentrated} random vectors. Further exploiting the fact that Gram matrices, of the type $G = X^\intercal X$ with $X=[x_1,\ldots,x_n]\in \mathbb{R}^{p\times n}$ and $x_i$ independent concentrated random vectors from a mixture model, behave asymptotically (as $n,p\to \infty$) as if the $x_i$ were drawn from a Gaussian mixture, suggests that DL representations of GAN-data can be fully described by their first two statistical moments for a wide range of standard classifiers. Our theoretical findings are validated by generating images with the BigGAN model and across different popular deep representation networks.

This article considers the problem of community detection in sparse dynamical graphs in which the community structure evolves over time. A fast spectral algorithm based on an extension of the Bethe-Hessian matrix is proposed, which benefits from the positive correlation in the class labels and in their temporal evolution and is designed to be applicable to any dynamical graph with a community structure. Under the dynamical degree-corrected stochastic block model, in the case of two classes of equal size, we demonstrate and support with extensive simulations that our proposed algorithm is capable of making non-trivial community reconstruction as soon as theoretically possible, thereby reaching the optimal detectability threshold and provably outperforming competing spectral methods.

This article characterizes the exact asymptotics of random Fourier feature (RFF) regression, in the realistic setting where the number of data samples $n$, their dimension $p$, and the dimension of feature space $N$ are all large and comparable. In this regime, the random RFF Gram matrix no longer converges to the well-known limiting Gaussian kernel matrix (as it does when $N \to \infty$ alone), but it still has a tractable behavior that is captured by our analysis. This analysis also provides accurate estimates of training and test regression errors for large $n,p,N$. Based on these estimates, a precise characterization of two qualitatively different phases of learning, including the phase transition between them, is provided; and the corresponding double descent test error curve is derived from this phase transition behavior. These results do not depend on strong assumptions on the data distribution, and they perfectly match empirical results on real-world data sets.

We propose a method for simultaneously detecting shared and unshared communities in heterogeneous multilayer weighted and undirected networks. The multilayer network is assumed to follow a generative probabilistic model that takes into account the similarities and dissimilarities between the communities. We make use of a variational Bayes approach for jointly inferring the shared and unshared hidden communities from multilayer network observations. We show that our approach outperforms state-of-the-art algorithms in detecting disparate (shared and private) communities on synthetic data as well as on real genome-wide fibroblast proliferation dataset.

This article improves over the recently proposed Bethe Hessian matrix for community detection on sparse graphs, assuming here a more realistic setting where node degrees are inhomogeneous. We notably show that the parametrization proposed in the seminal work on the Bethe Hessian clustering can be ameliorated with positive consequences on correct classification rates. Extensive simulations support our claims.

The recent work [1] shows that in the big data regime (i.e., numerous high dimensional data), the popular semi-supervised graph regularization, known as semi-supervised Laplacian regularization, fails to effectively extract information from unlabelled data. In response to this problem, we propose in this article an improved approach based on a simple yet fundamental update of the classical method. The effectiveness of the former is supported by both asymptotic results and simulations on finite data samples.

Logistic regression, one of the most popular machine learning binary classification methods, has been long believed to be unbiased. In this paper, we consider the "hard" classification problem of separating high dimensional Gaussian vectors, where the data dimension p and the sample size n are both large. Based on recent advances in random matrix theory (RMT) and high dimensional statistics, we evaluate the asymptotic distribution of the logistic regression classifier and consequently, provide the associated classification performance. This brings new insights into the internal mechanism of logistic regression classifier, including a possible bias in the separating hyperplane, as well as on practical issues such as hyper-parameter tuning, thereby opening the door to novel RMT-inspired improvements.

Based on recent random matrix advances in the analysis of kernel methods for classification and clustering, this paper proposes the study of large kernel methods for a wide class of random inputs, i.e., concentrated data, which are more generic than Gaussian mixtures. The concentration assumption is motivated by the fact that one can use generative models to design complex data structures, through Lipschitzally transformed concentrated vectors (e.g., Gaussian) which remain concentrated vectors. Applied to spectral clustering, we demonstrate that our theoretical findings closely match the behavior of large kernel matrices, when considering the fed-in data as CNN representations of GAN-generated images (i.e., concentrated vectors by design).

A wide range of machine learning and signal processing applications involve data discrimination through covariance matrices. A broad family of metrics, among which the Frobe-nius, Fisher, Bhattacharyya distances, as well as the Kullback-Leibler or Rényi divergences, are regularly exploited. Not being directly accessible, these metrics are usually assessed through empirical sample covariances. We show here that, for large dimensional data, these approximations lead to dramatically erroneous distance and divergence estimates.In this article, based on advanced random matrix considerations, we provide a novel and versatile consistent estimate for these covariance matrix distances and divergences. While theoretically developed for both large and numerous data, practical simulations demonstrate its large performance gains over the standard approach even for very small dimensions. A particular emphasis is made on the Fisher information metric and a concrete application to covariance-based spectral clustering is investigated.

Relying on recent advances in statistical estimation of covariance distances based on random matrix theory, this article proposes an improved covariance and precision matrix estimation for a wide family of metrics. The method is shown to largely outperform the sample covariance matrix estimate and to compete with state-of-the-art methods, while at the same time being computationally simpler and faster. Applications to linear and quadratic discriminant analyses also show significant gains, therefore suggesting practical interest to statistical machine learning.

Spectral clustering is one of the most popular, yet still incompletely understood, methods for community detection on graphs. This article studies spectral clustering based on the Bethe-Hessian matrix Hr= (r^2−1)In+D−rA for sparse heterogeneous graphs (following the degree-corrected stochastic block model) in a two-class setting. For a specific value r=ζ, clustering is shown to be insensitive to the degree heterogeneity. We then study the behavior of the informative eigenvector of H_ζ and, as a result, predict the clustering accuracy. The article concludes with an overview of the generalization to more than two classes along with extensive simulations on synthetic and real networks corroborating our findings.

We study in this article the asymptotic performance of spectral clustering with inner product kernel for Gaussian mixture models of high dimension with numerous samples. As is now classical in large dimensional spectral analysis, we establish a phase transition phenomenon by which a minimum distance between the class means and covariances is required for clustering to be possible from the dominant eigenvectors. Beyond this phase transition, we evaluate the asymptotic content of the dominant eigenvectors thus allowing for a full characterization of clustering performance. However, a surprising finding is that in some particular scenarios, the phase transition does not occur and clustering can be achieved irrespective of the class means and covariances. This is evidenced here in the case of the mixture of two Gaussian datasets having the same means and arbitrary difference between covariances.

This article provides a theoretical analysis of the asymptotic performance of a regression or classification task performed by a simple random neural network. This result is obtained by leveraging a new framework at the crossroads between random matrix theory and the concentration of measure theory. This approach is of utmost interest for neural network analysis at large in that it naturally dismisses the difficulty induced by the non-linear activation functions, so long that these are Lipschitz functions. As an application, we provide formulas for the limiting law of the random neural network output and compare them conclusively to those obtained practically on handwritten digits databases.

Random feature maps are ubiquitous in modern statistical machine learning, where they generalize random projections by means of powerful, yet often difficult to analyze nonlinear operators. In this paper we leverage the "concentration" phenomenon induced by random matrix theory to perform a spectral analysis on the Gram matrix of these random feature maps, here for Gaussian mixture models of simultaneously large dimension and size. Our results are instrumental to a deeper understanding on the interplay of the nonlinearity and the statistics of the data, thereby allowing for a better tuning of random feature-based techniques.

Understanding the learning dynamics of neural networks is one of the key issues for the improvement of optimization algorithms as well as for the theoretical comprehension of why deep neural nets work so well today. In this paper, we introduce a random matrix-based framework to analyze the learning dynamics of a single-layer linear network on a binary classification problem, for data of simultaneously large dimension and size, trained by gradient descent. Our results provide rich insights into common questions in neural nets, such as overfitting, early stopping and the initialization of training, thereby opening the door for future studies of more elaborate structures and models appearing in today’s neural networks.

The use of multiple antenna arrays in transmission and reception has become an integral part of modern wireless communications. To quantify the performance of such systems, the evaluation of bounds on the error probability of realistic finite length codewords is important. In this paper, we analyze the standard Gallager error bound for both constraints of maximum average power and maximum instantaneous power. Applying techniques from random matrix theory, we obtain analytic expressions of the error exponent when the length of the codeword increases to infinity at a fixed ratio with the antenna array dimensions. Analyzing its behavior at rates close to the ergodic rate, we find that the Gallager error bound becomes asymptotically close to an upper error bound obtained recently by Hoydis et al. 2015. We also obtain an expression for the Gallager exponent in the case when the codelength spans several Rayleigh fading blocks, hence taking into account the situation when the channel varies during each transmission.

This article proposes a performance analysis of kernel least squares support vector machines (LS-SVMs) based on a random matrix approach, in the regime where both the dimension of data p and their number n grow large at the same rate. Under a two-class Gaussian mixture model for the input data, we prove that the LS-SVM decision function is asymptotically normal with means and covariances shown to depend explicitly on the derivatives of the kernel function. This provides improved understanding along with new insights into the internal workings of SVM-type methods for large datasets.

This article proposes an original approach to the performance understanding of large dimensional neural networks. In this preliminary study, we study a single hidden layer feed-forward network with random input connections (also called extreme learning machine) which performs a simple regression task. By means of a new random matrix result, we prove that, as the size and cardinality of the input data and the number of neurons grow large, the network performance is asymptotically deterministic. This entails a better comprehension of the effects of the hyper-parameters (activation function, number of neurons, etc.) under this simple setting, thereby paving the path to the harnessing of more involved structures.

In this article, a new approach is proposed to study the performance of graph-based semi-supervised learning methods, under the assumptions that the dimension of data p and their number n grow large at the same rate and that the data arise from a Gaussian mixture model. Unlike small dimensional systems, the large dimensions allow for a Taylor expansion to linearize the weight (or kernel) matrix W, thereby providing in closed form the limiting performance of semi-supervised learning algorithms. This notably allows to predict the classification error rate as a function of the normalization parameters and of the choice of the kernel function. Despite the Gaussian assumption for the data, the theoretical findings match closely the performance achieved with real datasets, particularly here on the popular MNIST database.

In this paper, we study the physical layer multicasting to multiple co-channel groups in large-scale antenna systems. The users within each group are interested in a common message and different groups have distinct messages. In particular, we aim at designing the precoding vectors solving the so-called quality of service (QoS) and weighted max-min fairness (MMF) problems, assuming that the channel state information is available at the base station (BS). To solve both problems, the baseline approach exploits the semidefinite relaxation (SDR) technique. Considering a BS with $N$ antennas, the SDR complexity is more than $\mathcal {O}(N^{6})$ , which prevents its application in large-scale antenna systems. To overcome this issue, we present two new classes of algorithms that, not only have significantly lower computational complexity than existing solutions, but also largely outperform the SDR-based methods. Moreover, we present a novel duality between transformed versions of the QoS and the weighted MMF problems. The duality explicitly determines the solution to the weighted MMF problem given the solution to the QoS problem, and vice versa. Numerical results are used to validate the effectiveness of the proposed solutions and to make comparisons with existing alternatives under different operating conditions.

This article proposes a spectral analysis of dense random graphs generated by (a modified version of) the degree-corrected stochastic block model, for a setting where the inter block probabilities differ by O(n−) with n the number of nodes. We study a normalized version of the graph modularity matrix which is shown to be asymptotically well approximated by an analytically tractable (spiked) random matrix. The analysis of the latter allows for the precise evaluation of (i) the transition phase where clustering becomes asymptotically feasible and (ii) the alignment between the dominant eigenvectors and the block-wise canonical basis, thus enabling the estimation of mis-classification rates (prior to post-processing) in simple scenarios.

Recurrent neural networks, especially in their linear version, have provided many qualitative insights on their performance under different configurations. This article provides, through a novel random matrix framework, the quantitative counterpart of these performance results, specifically in the case of echo-state networks. Beyond mere insights, our approach conveys a deeper understanding on the core mechanism under play for both training and testing.

In this article, a study of the mean-square error (MSE) performance of linear echo-state neural networks is performed, both for training and testing tasks. Considering the realistic setting of noise present at the network nodes, we derive deterministic equivalents for the aforementioned MSE in the limit where the number of input data $T$ and network size $n$ both grow large. Specializing then the network connectivity matrix to specific random settings, we further obtain simple formulas that provide new insights on the performance of such networks.

This paper focuses on a large-scale multi-cell multi-user MIMO system in which L base stations (BSs) of N antennas each communicate with K single-antenna user equipments. We consider the design of the linear precoder that minimizes the total power consumption while ensuring target user rates. Three configurations with different degrees of cooperation among BSs are considered: the coordinated beamforming scheme (only channel state information is shared among BSs), the coordinated multipoint MIMO processing technology or network MIMO (channel state and data cooperation), and a single-cell beamforming scheme (only local channel state information is used for beamforming, while channel state cooperation is needed for power allocation). The analysis is conducted assuming that N and K$ grow large with a non trivial ratio K/N, and imperfect channel state information (modeled by the generic Gauss-Markov formulation form) is available at the BSs. Tools of random matrix theory are used to compute, in explicit form, deterministic approximations for: i) the parameters of the optimal precoder; ii) the powers needed to ensure target rates; and iii) the total transmit power. These results are instrumental to get further insight into the structure of the optimal precoders and also to reduce the implementation complexity in large-scale networks. Numerical results are used to validate the asymptotic analysis in the finite system regime and to make comparisons among the different configurations.

This work focuses on the downlink of a large-scale multi-cell multi-user MIMO system in which L base stations (BSs) of N antennas each communicate with KL single-antenna user equipments. We consider the design of the linear precoder that minimizes the total power consumption while ensuring target user rates. Two configurations with different degrees of cooperation among BSs are considered: the coordinated beamforming scheme (only channel state information is shared between BSs) and the coordinated multipoint MIMO technology (channel state and data cooperation). The analysis is conducted assuming that N and K grow large with a non trivial ratio K/N and imperfect channel state information is available at the BSs. In both configurations, tools of random matrix theory are used to compute, often in closed form, deterministic approximations for: the parameters of the optimal precoder; the powers needed to ensure target rates; and the total transmit power. These results are instrumental to get further insights into the structure of the optimal precoder and also to reduce the complexity of its implementation in large-scale networks. Numerical results are used to validate the asymptotic analysis in the finite system regime and to make comparisons among the two different configurations.

This paper lies in the lineage of recent works studying the asymptotic behaviour of robust-scatter estimators in the case where the number of observations and the dimension of the population covariance matrix grow at infinity with the same pace. In particular, we analyze the fluctuations of bilinear forms of the robust shrinkage estimator of covariance matrix. We show that this result can be leveraged in order to improve the design of robust detection methods. As an example, we provide an improved generalized likelihood ratio based detector which combines robustness to impulsive observations and optimality across the shrinkage parameter, the optimality being considered for the false alarm regulation.

Building on recent results in the random matrix analysis of robust estimators of scatter, we show that a certain class of such estimators obtained from samples containing outliers behaves similar to a well-known random matrix model in the limiting regime where both the population and sample sizes grow to infinity at the same speed. This result allows us to understand the structure of such estimators when a certain fraction of the samples is corrupted by outliers and, in particular, to derive their asymptotic eigenvalue distributions. This analysis is a first step towards an improved usage of robust estimation methods under the presence of outliers when the number of independent observations is not too large compared to the size of the population.

A distributed hypothesis testing (HT) problem is considered, comprising two nodes and a unidirectional communication link. The receiving node is required to make a decision as to the probability distribution in effect. A binning process is used in order to minimize the probability of error, resulting in a new achievable error-exponent. A sub-class of HT problems with general hypotheses is defined, which contains many interesting and relevant problems. The advantage of the binning strategy in comparison to the non-binning approach is demonstrated by means of a binary symmetric example.

In this letter, we propose an alternative proof for the uniqueness of Maronna's M-estimator of scatter for N vector observations y1, ..., yN∈ Rmunder a mild constraint of linear independence of any subset of m of these vectors. This entails in particular almost sure uniqueness for random vectors yi with a density as long as N > m. This approach allows to establish further relations that demonstrate that a properly normalized Tyler's M-estimator of scatter can be considered as a limit of Maronna's M-estimator. More precisely, the contribution is to show that each M-estimator, verifying some mild conditions, converges towards a particular Tyler's M-estimator. These results find important implications in recent works on the large dimensional (random matrix) regime of robust M-estimation.

The second-order coding rate of the multiple-input multiple-output (MIMO) quasi-static Rayleigh fading channel is studied. We tackle this problem via an information-spectrum approach and statistical bounds based on recent random matrix theory techniques. We derive a central limit theorem (CLT) to analyze the information density in the regime where the block length n and the number of transmit and receive antennas K and N, respectively, grow simultaneously large. This result leads to the characterization of closed-form upper and lower bounds on the optimal average error probability when the coding rate is within O(1/√(nK)) of the asymptotic capacity.

In a large network of agents, we consider a distributed convex optimization problem where each agent has a private convex cost function and a set of local variables. We provide an algorithm to carry out a multi-area decentralized optimization in an asynchronous fashion, obtained by applying random Gauss-Seidel iterations on the Douglas-Rachford splitting operator. As an application, a direct-current linear optimal power flow model is implemented and simulations results confirm the convergence of the proposed algorithm.

This paper studies the limiting behavior of a class of robust population covariance matrix estimators, originally due to Maronna in 1976, in the regime where both the number of available samples and the population size grow large. Using tools from random matrix theory, we prove that, for sample vectors made of independent entries having some moment conditions, the difference between the sample covariance matrix and (a scaled version of) such robust estimator tends to zero in spectral norm, almost surely. This result can be applied to various statistical methods arising from random matrix theory that can be made robust without altering their first order behavior.

An original interface between robust estimation theory and random matrix theory for the estimation of population covariance matrices is proposed. Consider a random vector x = ANy ∈ CNwith y ∈ CMmade of M ≥ N independent entries, E[y] = 0, and E[yy*] = IN. It is shown that a class of robust estimators ĈNof CN= ANA*N, obtained from n independent copies of x, is (N, n)-consistent with the traditional sample covariance matrix r̂Nin the sense that ∥ĈN- αr̂N∥ → 0 in spectral norm for some α > 0, almost surely, as N, n → ∞ with N/n and M/N bounded. This result, in general not valid in the fixed N regime, is used to propose improved subspace estimation techniques, among which an enhanced direction-of-arrival estimator called robust G-MUSIC.

In this paper, we study the performance of regularized channel inversion precoding in MISO broadcast channels with confidential messages under imperfect channel state information at the transmitter (CSIT). We obtain an approximation for the achievable secrecy sum-rate which is almost surely exact as the number of transmit antennas and the number of users grow to infinity in a fixed ratio. Simulations prove this anaylsis accurate even for finite-size systems. For FDD systems, we determine how the CSIT error must scale with the SNR, and we derive the number of feedback bits required to ensure a constant high-SNR rate gap to the case with perfect CSIT. For TDD systems, we study the optimum amount of channel training that maximizes the high-SNR secrecy sum-rate.

Most statistical inference methods for array processing assume an array of size N fixed and a number of snapshots T large. In addition, many works are based on the assumption of a white noise model. These two assumptions are increasingly less realistic in modern systems where N and T are usually both large, and where the noise data can be correlated either across successive observations or across the sensor antennas. In this paper an approach to handle this kind of scenario is presented. New algorithms for source number estimation, power estimation, and localization by a sensor array under noise with unknown correlation model are proposed. The results fundamentally rely on recent advances in small rank perturbations of large dimensional random matrices.

We study the second-order coding rate of the multiple-input multiple-output (MIMO) Rayleigh block-fading channel via statistical bounds from information spectrum methods and random matrix theory. Based on an asymptotic analysis of the mutual information density which considers the simultaneous growth of the block length n and the number of transmit and receive antennas K and N, we derive closed-form upper and lower bounds on the optimal average error probability when the code rate is within O(1/√nK) of the asymptotic capacity. A Gaussian approximation is then used to establish an upper bound on the error probability for arbitrary code rates which is shown by simulations to be accurate for small N, K, and n.

In this paper, we study the performance of regularized channel inversion (RCI) precoding in large MISO broadcast channels with confidential messages (BCC). We obtain a deterministic approximation for the achievable secrecy sum-rate which is almost surely exact as the number of transmit antennas M and the number of users K grow to infinity in a fixed ratio β=K/M. We derive the optimal regularization parameter ξ and the optimal network load β that maximize the per-antenna secrecy sum-rate. We then propose a linear precoder based on RCI and power reduction (RCI-PR) that significantly increases the high-SNR secrecy sum-rate for 1<;β<;2. Our proposed precoder achieves a per-user secrecy rate which has the same high-SNR scaling factor as both the following upper bounds: (i) the rate of the optimum RCI precoder without secrecy requirements, and (ii) the secrecy capacity of a single-user system without interference. Furthermore, we obtain a deterministic approximation for the secrecy sum-rate achievable by RCI precoding in the presence of channel state information (CSI) error. We also analyze the performance of our proposed RCI-PR precoder with CSI error, and we determine how the error must scale with the SNR in order to maintain a given rate gap to the case with perfect CSI.

In this paper, the joint fluctuations of the extreme eigenvalues and eigenvectors of a large dimensional sample covariance matrix are analyzed when the associated population covariance matrix is a finite-rank perturbation of the identity matrix, corresponding to the so-called spiked model in random matrix theory. The asymptotic fluctuations, as the matrix size grows large, are shown to be intimately linked with matrices from the Gaussian unitary ensemble. When the spiked population eigenvalues have unit multiplicity, the fluctuations follow a central limit theorem. This result is used to develop an original framework for the detection and diagnosis of local failures in large sensor networks, for known or unknown failure magnitude.

In this paper, the problem of fast point-to-point multiple-input-multiple-output channel mutual information estimation is addressed, in the situation where the receiver undergoes unknown colored interference, whereas the channel with the transmitter is perfectly known. The considered scenario assumes that the estimation is based on a few channel use observations during a short sensing period. Using large dimensional random matrix theory, an estimator referred to as G-estimator is derived. This estimator is proved to be consistent as the number of antennas and observations grow large and its asymptotic performance is analyzed. In particular, the G-estimator satisfies a central limit theorem with asymptotic Gaussian fluctuations. Simulations are provided which strongly support the theoretical results, even for small system dimensions.

This paper provides a central limit theorem for a consistent estimator of population eigenvalues with large multiplicities based on sample covariance matrices. The focus is on limited sample size situations, whereby the number of available observations is comparable in magnitude to the observation dimension. An exact expression as well as an empirical, asymptotically accurate, approximation of the limiting variance is derived. Simulations are performed that corroborate the theoretical claims.

This article proposes a large system approximation of the ergodic sum-rate (SR) for cellular multi-user multiple-input multiple-output uplink systems. The considered system has various degrees of freedom, such as clusters of base stations (BSs) performing cooperative multi-point processing, randomly distributed user terminals (UTs), and supports arbitrarily configurable antenna gain patterns at the BSs. The approximation is provably tight in the limiting case of a large number of single antenna UTs and antennas at the BSs. Simulation results suggest that the asymptotic analysis is accurate for small system dimensions. Our deterministic SR approximation result is applied to numerically study and optimize the effects of antenna tilting in an exemplary sectorized 3D small cell network topology. Significant SR gains are observed with optimal tilt angles and we provide new insights on the optimal parameterization of cellular networks, along with a discussion of several non-trivial effects.

In the context of multidimensional signals, the linear Wiener receiver is frequently encountered in wireless communication and in array processing; it is in fact the linear receiver that achieves the lowest level of interference. In this contribution, we focus on the study of the associated Signal-to-interference plus noise ratio (SINR) at its output in the context of Ricean multiple-input multiple-output (MIMO) channels. The case of Ricean channels, which induces non-centered random variables, can be encountered in several practical environments and has not been studied so far, as it raises substantial technical issues. With the help of large random matrix theory, which has shown to be fruitful to successfully address several problems in wireless communications, we study the behaviour of the SINR, together with its fluctuations via a central limit theorem. As realistic models also involve non-Gaussian random variables, we relax the Gaussian assumption. This results in an extra term involving the fourth cumulant in the expression of the variance.

This paper provides a novel central limit theorem (CLT) for the information density of the MIMO Rayleigh fading channel under white Gaussian inputs, when the data blocklength n and the number of transmit and receive antennas K and N, respectively, are large but of similar order of magnitude. This CLT is used to derive closed-form upper bounds on the error probability via an input-constrained version of Feinstein's lemma by Polyanskiy et al. and the second-order approximation of the coding rate. Numerical evaluations suggest that the normal approximation is tight for reasonably small values of n, K, N.

In this article, we investigate the competitive interaction between electrical vehicles or hybrid oil-electricity vehicles in a Cournot market consisting of electricity transactions to or from an underlying electricity distribution network. We provide a mean field game formulation for this competition, and introduce the set of fundamental differential equations ruling the behavior of the vehicles at the feedback Nash equilibrium, referred here to as the mean field equilibrium. This framework allows for a consistent analysis of the evolution of the price of electricity as well as of the instantaneous electricity demand in the power grid. Simulations precisely quantify those parameters and suggest that significant reduction of the daily electricity peak demand can be achieved by appropriate electricity pricing.

In this work, we study the performance of random isometric precoders over quasi-static and correlated fading channels. We derive deterministic approximations of the mutual information and the signal-to-interference-plus-noise ratio (SINR) at the output of the minimum-mean-square-error (MMSE) receiver and provide simple provably converging fixed-point algorithms for their computation. Although these approximations are only proven exact in the asymptotic regime with infinitely many antennas at the transmitters and receivers, simulations suggest that they closely match the performance of small-dimensional systems. We exemplarily apply our results to the performance analysis of multi-cellular communication systems, multiple-input multiple-output multiple-access channels (MIMO-MAC), and MIMO interference channels. The mathematical analysis is based on the Stieltjes transform method. This enables the derivation of deterministic equivalents of functionals of large-dimensional random matrices. In contrast to previous works, our analysis does not rely on arguments from free probability theory which enables the consideration of random matrix models for which asymptotic freeness does not hold. Thus, the results of this work are also a novel contribution to the field of random matrix theory and applicable to a wide spectrum of practical systems.

In this paper, we study the sum rate performance of zero-forcing (ZF) and regularized ZF (RZF) precoding in large MISO broadcast systems under the assumptions of imperfect channel state information at the transmitter and per-user channel transmit correlation. Our analysis assumes that the number of transmit antennas M and the number of single-antenna users K are large while their ratio remains bounded. We derive deterministic approximations of the empirical signal-to-interference plus noise ratio (SINR) at the receivers, which are tight as M, K → ∞. In the course of this derivation, the per-user channel correlation model requires the development of a novel deterministic equivalent of the empirical Stieltjes transform of large dimensional random matrices with generalized variance profile. The deterministic SINR approximations enable us to solve various practical optimization problems. Under sum rate maximization, we derive 1) for RZF the optimal regularization parameter; 2) for ZF the optimal number of users; 3) for ZF and RZF the optimal power allocation scheme; and 4) the optimal amount of feedback in large FDD/TDD multiuser systems. Numerical simulations suggest that the deterministic approximations are accurate even for small M, K.

In this paper, we address the problem of fast point-to-point channel capacity estimation in the case where the receiver undergoes unknown interference from multiple sources, whereas the channel with the transmitter is perfectly known. For this particular context, we propose a fast estimator for the capacity estimation, and compare its performance with that of the traditional methods. More precisely, we analyse the fluctuations of the traditional and proposed techniques and prove that their behaviors can be approximated by Gaussian random variables for which we derive the variances.

This article provides a central limit theorem for a consistent estimator of the population eigenvalues of a class of sample covariance matrices. An exact expression as well as an empirical and asymptotically accurate approximation of the limiting variance is also derived. These results are applied in a cognitive radio context featuring an orthogonal-CDMA primary network and a secondary network whose objective is to maximise the coverage of secondary transmissions under low probability of interference with primary users.

We consider a general wireless channel model for different types of code-division multiple access (CDMA) and space-division multiple-access (SDMA) systems with isometric random signature/precoding matrices over frequency-selective and flat fading channels. We derive deterministic approximations of the Stieltjes transform, the mutual information and the signal-to-interference-plus-noise ratio (SINR) at the output of the minimum-mean-square-error (MMSE) receiver and provide a simple fixed-point algorithm for their computation, which is proved to converge. The deterministic approximations are asymptotically tight, almost surely, but shown by simulations to be very accurate for even small system dimensions. Our analysis requires neither arguments from free probability theory nor the asymptotic freeness or the convergence of the spectral distribution of the involved matrices. The results presented in this work are, therefore, also a novel contribution to the field of random matrix theory and might be useful to further applications involving isometric random matrices.

This paper considers the MISO broadcast channel with different spatial correlations of the user vector channels. The base station implements regularized zero-forcing (RZF) precoding based on an imperfect channel estimation. We derive a deterministic equivalent of the signal-to-interference plus noise ratio (SINR) by applying novel results from the field of large dimensional random matrices. Based on this deterministic equivalent, we compute the sum rate maximizing RZF precoder which is given in closed form for independent and identically distributed channels. Simulations show that the accuracy of the approximated SINR extends well into finite dimensions.

In this article, novel deterministic equivalents for the Stieltjes transform and the Shannon transform of a class of large dimensional random matrices are provided. These results are used to characterize the ergodic rate region of multiple antenna multiple access channels, when each point-to-point propagation channel is modelled according to the Kronecker model. Specifically, an approximation of all rates achieved within the ergodic rate region is derived and an approximation of the linear precoders that achieve the boundary of the rate region as well as an iterative water-filling algorithm to obtain these precoders are provided. An original feature of this work is that the proposed deterministic equivalents are proved valid even for strong correlation patterns at both communication sides. The above results are validated by Monte Carlo simulations.

In this paper, a new method is introduced to blindly estimate the transmit power of multiple signal sources in multiantenna fading channels, when the number of sensing devices and the number of available samples are sufficiently large compared to the number of sources. Recent advances in the field of large dimensional random matrix theory are used that result in a simple and computationally efficient consistent estimator of the power of each source. A criterion to determine the minimum number of sensors and the minimum number of samples required to achieve source separation is then introduced. Simulations are performed that corroborate the theoretical claims and show that the proposed power estimator largely outperforms alternative power inference techniques.

This paper considers a large multi-user time-division duplex (TDD) system, where the base station (BS) acquires channel state information via pilot signaling from the users. In the downlink the BS employs zero-forcing (ZF) and regularized zero-forcing (RZF) precoding. We derive the optimal sum rate maximizing amount of channel training using sum rate approximations from the large system analysis of MISO downlink channels under (R)ZF precoding. Moreover, in the regime of high signal-to-noise ratio (SNR), we derive approximate solutions of the optimal amount of training for both schemes that are of closed-form. By comparing the two schemes, we find that RZF requires less training than ZF, but the training interval of both schemes is equal for asymptotically high SNR. Furthermore, simulations are carried out which demonstrate the accuracy of our approximate solutions.

In this paper, a deterministic approximation for the rate region of multiple access channels is provided when both the base station and the users have large numbers of antennas, and when the transmission bandwidth is divided into several independent subbands. An explicit formulation is also given for the transmit covariance matrices, at each frequency, that reach the boundary of the rate region. From the compact expression of these matrices, suboptimal iterative algorithms emerge that allow the multiple access users to derive autonomously the transmit covariance matrices. This comes at the sole expense of a small amount of signalling overhead, which is constant irrespectively of the number of antennas. Simulations confirm the validity of the theoretical derivations and suggest rather good behavior of the suboptimal self-organization algorithms.

This paper introduces a new method to estimate the power transmitted by multiple signal sources, when the number of sensing devices and the available samples are sufficiently large compared to the number of sources. This work makes use of recent advances in the field of random matrix theory that prove more efficient than previous “moment-based” approaches to the problem of multi-source power detection. Simulations are performed which corroborate the theoretical claims.

In this paper, alpha-OFDM, a generalization of the OFDM modulation, is proposed to enhance the outage capacity of bursty transmissions. This new flexible modulation scheme is easily implemented and only requires a symbol rotation of angle alpha after the IDFT stage. The induced rotation slides the DFT window and provides frequency diversity in block fading channels. Interestingly, the results show a substantial gain in terms of outage capacity and BER in comparison with classical OFDM modulation schemes. The framework is extended to multiuser/multi-antenna OFDM based standards. Simulations, in the context of 3GPP LTE, called hereafter alpha-LTE, sustain our theoretical claims.

In this paper, we provide a Bayesian learning process for cognitive devices. In particular we focus on the case of signal detection as an explanatory example to the learning framework. Under any prior state of knowledge on the communication channel, an information theoretic criterion is presented to decide if informative data is present in a noisy wireless MIMO communication. We detail the particular cases of knowledge, or absence of knowledge at the receiver, of (i) the number of transmit antennas and (ii) the effective noise power. The provided method is instrumental to embed intelligence into the wireless device and gives birth to a novel Bayesian signal detector which is compared to the classical power detector. Simulations corroborate the theoretical results and quantify the gain achieved by the proposed Bayesian framework.

In this paper we consider the MIMO broadcast channel with antenna correlation at the transmitter and receiver. We derive the theoretical sum rate of systems with a large number of antennas for zero-forcing and regularized zero-forcing precoders. Particularly, we apply the results to volume-limited devices where the correlation originates from a dense antenna packing. Throughout this contribution we make extensive use of recent tools from random matrix theory. Simulations confirm the theoretical claims and also indicate that in most scenarios the asymptotic derivations applied to a finite number of users give good approximations of the true ergodic sum rate.

In this paper, a new multicell OFDM blind power detection method is proposed. Relying on recent results of free deconvolution, our algorithm enables the terminal to count the number of surrounding base stations and to determine the power received from each of them, based on a limited number of snapshots. This is in sharp contrast with classical asymptotic blind techniques. A theoretical analysis is proposed to study the impact of frequency selectivity and the number of receive/transmit antennas. Simulations are provided to sustain the theoretical claims and are compared against classical techniques.