arXiv subjects

Stefano Rini

Publications and source records attributed to Stefano Rini.

At least 37 records · Page 2Linked to original sources

arXiv2023

Harmonic Retrieval Using Weighted Lifted-Structure Low-Rank Matrix Completion

In this paper, we investigate the problem of recovering the frequency components of a mixture of $K$ complex sinusoids from a random subset of $N$ equally-spaced time-domain samples. Because of the random subset, the samples are effectively non-uniform. Besides, the frequency values of each of the $K$ complex sinusoids are assumed to vary continuously within a given range. For this problem, we propose a two-step strategy: (i) we first lift the incomplete set of uniform samples (unavailable samples are treated as missing data) into a structured matrix with missing entries, which is potentially low-rank; then (ii) we complete the matrix using a weighted nuclear minimization problem. We call the method a \emph{ weighted lifted-structured (WLi) low-rank matrix recovery}. Our approach can be applied to a range of matrix structures such as Hankel and double-Hankel, among others, and provides improvement over the unweighted existing schemes such as EMaC and DEMaC. We provide theoretical guarantees for the proposed method, as well as numerical simulations in both noiseless and noisy settings. Both the theoretical and the numerical results confirm the superiority of the proposed approach.

eess.SP↗

arXiv2023

Coding for the unsourced A-channel with erasures: the linked loop code

The A-channel is a noiseless multiple access channel in which users simultaneously transmit Q-ary symbols and the receiver observes the set of transmitted symbols, but not their multiplicities. An A-channel is said to be unsourced if, additionally, users transmissions are encoded across time using a common codebook and decoding of the transmitted messages is done without regard to the identities of the active users. An interesting variant of the unsourced A-channel is the unsourced A-channel with erasures (UACE), in which transmitted symbols are erased with a given independent and identically distributed probability. In this paper, we focus on designing a code that enables a list of transmitted codewords to be recovered despite the erasures of some of the transmitted symbols. To this end, we propose the linked-loop code (LLC), which uses parity bits to link each symbol to the previous M symbols in a tail-biting manner, i.e., the first symbols of the transmission are linked to the last ones. The decoding process occurs in two phases: the first phase decodes the codewords that do not suffer from any erasures, and the second phase attempts to recover the erased symbols using the available parities. We compare the performance of the LLC over the UACE with other codes in the literature and argue for the effectiveness of the construction. Our motivation for studying the UACE comes from its relevance in machine-type communication and coded compressed sensing.

cs.IT↗

arXiv2023

How to Attain Communication-Efficient DNN Training? Convert, Compress, Correct

This paper introduces CO3 -- an algorithm for communication-efficient federated Deep Neural Network (DNN) training. CO3 takes its name from three processing applied which reduce the communication load when transmitting the local DNN gradients from the remote users to the Parameter Server. Namely: (i) gradient quantization through floating-point conversion, (ii) lossless compression of the quantized gradient, and (iii) quantization error correction. We carefully design each of the steps above to assure good training performance under a constraint on the communication rate. In particular, in steps (i) and (ii), we adopt the assumption that DNN gradients are distributed according to a generalized normal distribution, which is validated numerically in the paper. For step (iii), we utilize an error feedback with memory decay mechanism to correct the quantization error introduced in step (i). We argue that the memory decay coefficient, similarly to the learning rate, can be optimally tuned to improve convergence. A rigorous convergence analysis of the proposed CO3 with SGD is provided. Moreover, with extensive simulations, we show that CO3 offers improved performance when compared with existing gradient compression schemes in the literature which employ sketching and non-uniform quantization of the local gradients.

cs.LG↗

arXiv2023

M22: A Communication-Efficient Algorithm for Federated Learning Inspired by Rate-Distortion

In federated learning (FL), the communication constraint between the remote learners and the Parameter Server (PS) is a crucial bottleneck. For this reason, model updates must be compressed so as to minimize the loss in accuracy resulting from the communication constraint. This paper proposes ``\emph{${\bf M}$-magnitude weighted $L_{\bf 2}$ distortion + $\bf 2$ degrees of freedom''} (M22) algorithm, a rate-distortion inspired approach to gradient compression for federated training of deep neural networks (DNNs). In particular, we propose a family of distortion measures between the original gradient and the reconstruction we referred to as ``$M$-magnitude weighted $L_2$'' distortion, and we assume that gradient updates follow an i.i.d. distribution -- generalized normal or Weibull, which have two degrees of freedom. In both the distortion measure and the gradient, there is one free parameter for each that can be fitted as a function of the iteration number. Given a choice of gradient distribution and distortion measure, we design the quantizer minimizing the expected distortion in gradient reconstruction. To measure the gradient compression performance under a communication constraint, we define the \emph{per-bit accuracy} as the optimal improvement in accuracy that one bit of communication brings to the centralized model over the training period. Using this performance measure, we systematically benchmark the choice of gradient distribution and distortion measure. We provide substantial insights on the role of these choices and argue that significant performance improvements can be attained using such a rate-distortion inspired compressor.

cs.LG↗

arXiv2022

Empirical Risk Minimization with Relative Entropy Regularization: Optimality and Sensitivity Analysis

The optimality and sensitivity of the empirical risk minimization problem with relative entropy regularization (ERM-RER) are investigated for the case in which the reference is a sigma-finite measure instead of a probability measure. This generalization allows for a larger degree of flexibility in the incorporation of prior knowledge over the set of models. In this setting, the interplay of the regularization parameter, the reference measure, the risk function, and the empirical risk induced by the solution of the ERM-RER problem is characterized. This characterization yields necessary and sufficient conditions for the existence of a regularization parameter that achieves an arbitrarily small empirical risk with arbitrarily high probability. The sensitivity of the expected empirical risk to deviations from the solution of the ERM-RER problem is studied. The sensitivity is then used to provide upper and lower bounds on the expected empirical risk. Moreover, it is shown that the expectation of the sensitivity is upper bounded, up to a constant factor, by the square root of the lautum information between the models and the datasets.

cs.LG↗

arXiv2022

A Perspective on Neural Capacity Estimation: Viability and Reliability

Recently, several methods have been proposed for estimating the mutual information from sample data using deep neural networks. These estimators ar referred to as neural mutual information estimation (NMIE)s. NMIEs differ from other approaches as they are data-driven estimators. As such, they have the potential to perform well on a large class of capacity problems. In order to test the performance across various NMIEs, it is desirable to establish a benchmark encompassing the different challenges of capacity estimation. This is the objective of this paper. In particular, we consider three scenarios for benchmarking:i the classic AWGN channel, ii channels continuous inputs optical intensity and peak-power constrained AWGN channel iii channels with a discrete output, i.e., Poisson channel. We also consider the extension to the multi-terminal case with iv the AWGN and optical MAC models. We argue that benchmarking a certain NMIE across these four scenarios provides a substantive test of performance. In this paper we study the performance of mutual information neural estimator (MINE), smoothed mutual information lower-bound estimator (SMILE), and directed information neural estimator (DINE). and provide insights on the performance of other methods as well. To summarize our benchmarking results, MINE provides the most reliable performance.

cs.IT↗

arXiv2022

Sharp asymptotics on the compression of two-layer neural networks

In this paper, we study the compression of a target two-layer neural network with N nodes into a compressed network with M<N nodes. More precisely, we consider the setting in which the weights of the target network are i.i.d. sub-Gaussian, and we minimize the population L_2 loss between the outputs of the target and of the compressed network, under the assumption of Gaussian inputs. By using tools from high-dimensional probability, we show that this non-convex problem can be simplified when the target network is sufficiently over-parameterized, and provide the error rate of this approximation as a function of the input dimension and N. In this mean-field limit, the simplified objective, as well as the optimal weights of the compressed network, does not depend on the realization of the target network, but only on expected scaling factors. Furthermore, for networks with ReLU activation, we conjecture that the optimum of the simplified optimization problem is achieved by taking weights on the Equiangular Tight Frame (ETF), while the scaling of the weights and the orientation of the ETF depend on the parameters of the target network. Numerical evidence is provided to support this conjecture.

cs.IT↗

arXiv2022

Coded Demixing for Unsourced Random Access

Unsourced random access (URA) is a recently proposed multiple access paradigm tailored to the uplink channel of machine-type communication networks. By exploiting a strong connection between URA and compressed sensing, the massive multiple access problem may be cast as a compressed sensing (CS) problem, albeit one in exceedingly large dimensions. To efficiently handle the dimensionality of the problem, coded compressed sensing (CCS) has emerged as a pragmatic signal processing tool that, when applied to URA, offers good performance at low complexity. While CCS is effective at recovering a signal that is sparse with respect to a single basis, it is unable to jointly recover signals that are sparse with respect to separate bases. In this article, the CCS framework is extended to the demixing setting, yielding a novel technique called coded demixing. A generalized framework for coded demixing is presented and a low-complexity recovery algorithm based on approximate message passing (AMP) is developed. Coded demixing is applied to heterogeneous multi-class URA networks and traditional single-class networks. Its performance is analyzed and numerical simulations are presented to highlight the benefits of coded demixing.

cs.IT↗

arXiv2022

Lossy Gradient Compression: How Much Accuracy Can One Bit Buy?

In federated learning (FL), a global model is trained at a Parameter Server (PS) by aggregating model updates obtained from multiple remote learners. Generally, the communication between the remote users and the PS is rate-limited, while the transmission from the PS to the remote users are unconstrained. The FL setting gives rise to the distributed learning scenario in which the updates from the remote learners have to be compressed so as to meet communication rate constraints in the uplink transmission toward the PS. For this problem, one wishes to compress the model updates so as to minimize the loss in accuracy resulting from the compression error. In this paper, we take a rate-distortion approach to address the compressor design problem for the distributed training of deep neural networks (DNNs). In particular, we define a measure of the compression performance under communication-rate constraints -- the \emph{per-bit accuracy} -- which addresses the ultimate improvement of accuracy that a bit of communication brings to the centralized model. In order to maximize the per-bit accuracy, we consider modeling the DNN gradient updates at remote learners as a generalized normal distribution. Under this assumption on the DNN gradient distribution, we propose a class of distortion measures to aid the design of quantizers for the compression of the model updates. We argue that this family of distortion measures, which we refer to as "$M$-magnitude weighted $L_2$" norm, captures the practitioner's intuition in the choice of gradient compressor. Numerical simulations are provided to validate the proposed approach for the CIFAR-10 dataset.

cs.LG↗

arXiv2022

Neural Capacity Estimators: How Reliable Are They?

Recently, several methods have been proposed for estimating the mutual information from sample data using deep neural networks and without the knowing closed form distribution of the data. This class of estimators is referred to as neural mutual information estimators. Although very promising, such techniques have yet to be rigorously bench-marked so as to establish their efficacy, ease of implementation, and stability for capacity estimation which is joint maximization frame-work. In this paper, we compare the different techniques proposed in the literature for estimating capacity and provide a practitioner perspective on their effectiveness. In particular, we study the performance of mutual information neural estimator (MINE), smoothed mutual information lower-bound estimator (SMILE), and directed information neural estimator (DINE) and provide insights on InfoNCE. We evaluated these algorithms in terms of their ability to learn the input distributions that are capacity approaching for the AWGN channel, the optical intensity channel, and peak power-constrained AWGN channel. For both scenarios, we provide insightful comments on various aspects of the training process, such as stability, sensitivity to initialization.

cs.IT↗

arXiv2022

Convert, compress, correct: Three steps toward communication-efficient DNN training

In this paper, we introduce a novel algorithm, $\mathsf{CO}_3$, for communication-efficiency distributed Deep Neural Network (DNN) training. $\mathsf{CO}_3$ is a joint training/communication protocol, which encompasses three processing steps for the network gradients: (i) quantization through floating-point conversion, (ii) lossless compression, and (iii) error correction. These three components are crucial in the implementation of distributed DNN training over rate-constrained links. The interplay of these three steps in processing the DNN gradients is carefully balanced to yield a robust and high-performance scheme. The performance of the proposed scheme is investigated through numerical evaluations over CIFAR-10.

cs.LG↗

arXiv2022

Compressibility Measures for Affinely Singular Random Vectors

There are several ways to measure the compressibility of a random measure; they include general approaches such as using the rate-distortion curve, as well as more specific notions, such as the Renyi information dimension (RID). The RID parameter indicates the concentration of the measure around lower-dimensional subsets of the space. While the evaluation of such compressibility parameters is well-studied for continuous and discrete measures, the case of discrete-continuous measures is quite subtle. In this paper, we focus on a class of multi-dimensional random measures that have singularities on affine lower-dimensional subsets. This class of distributions naturally arises when considering linear transformation of component-wise independent discrete-continuous random variables. To measure the compressibility of such distributions, we introduce the new notion of dimensional-rate bias (DRB) which is closely related to the entropy and differential entropy in discrete and continuous cases, respectively. Similar to entropy and differential entropy, DRB is useful in evaluating the mutual information between distributions of the aforementioned type. Besides the DRB, we also evaluate the the RID of these distributions. We further provide an upper-bound for the RID of multi-dimensional random measures that are obtained by Lipschitz functions of component-wise independent discrete-continuous random variables ($\mathbf{X}$). The upper-bound is shown to be achievable when the Lipschitz function is $A \mathbf{X}$, where $A$ satisfies {\changed$\spark({A_{m\times n}}) = m+1$} (e.g., Vandermonde matrices). When considering discrete-domain moving-average processes with non-Gaussian excitation noise, the above results allow us to evaluate the block-average RID and DRB, as well as to determine a relationship between these parameters and other existing compressibility measures.

cs.IT↗

arXiv2022

Two-snapshot DOA Estimation via Hankel-structured Matrix Completion

In this paper, we study the problem of estimating the direction of arrival (DOA) using a sparsely sampled uniform linear array (ULA). Based on an initial incomplete ULA measurement, our strategy is to choose a sparse subset of array elements for measuring the next snapshot. Then, we use a Hankel-structured matrix completion to interpolate for the missing ULA measurements. Finally, the source DOAs are estimated using a subspace method such as Prony on the fully recovered ULA. We theoretically provide a sufficient bound for the number of required samples (array elements) for perfect recovery. The numerical comparisons of the proposed method with existing techniques such as atomic-norm minimization and off-the-grid approaches confirm the superiority of the proposed method.

eess.SP↗

arXiv2021

DNN gradient lossless compression: Can GenNorm be the answer?

In this paper, the problem of optimal gradient lossless compression in Deep Neural Network (DNN) training is considered. Gradient compression is relevant in many distributed DNN training scenarios, including the recently popular federated learning (FL) scenario in which each remote users are connected to the parameter server (PS) through a noiseless but rate limited channel. In distributed DNN training, if the underlying gradient distribution is available, classical lossless compression approaches can be used to reduce the number of bits required for communicating the gradient entries. Mean field analysis has suggested that gradient updates can be considered as independent random variables, while Laplace approximation can be used to argue that gradient has a distribution approximating the normal (Norm) distribution in some regimes. In this paper we argue that, for some networks of practical interest, the gradient entries can be well modelled as having a generalized normal (GenNorm) distribution. We provide numerical evaluations to validate that the hypothesis GenNorm modelling provides a more accurate prediction of the DNN gradient tail distribution. Additionally, this modeling choice provides concrete improvement in terms of lossless compression of the gradients when applying classical fix-to-variable lossless coding algorithms, such as Huffman coding, to the quantized gradient updates. This latter results indeed provides an effective compression strategy with low memory and computational complexity that has great practical relevance in distributed DNN training scenarios.

cs.LG↗

arXiv2021

Speeding-Up Back-Propagation in DNN: Approximate Outer Product with Memory

In this paper, an algorithm for approximate evaluation of back-propagation in DNN training is considered, which we term Approximate Outer Product Gradient Descent with Memory (Mem-AOP-GD). The Mem-AOP-GD algorithm implements an approximation of the stochastic gradient descent by considering only a subset of the outer products involved in the matrix multiplications that encompass backpropagation. In order to correct for the inherent bias in this approximation, the algorithm retains in memory an accumulation of the outer products that are not used in the approximation. We investigate the performance of the proposed algorithm in terms of DNN training loss under two design parameters: (i) the number of outer products used for the approximation, and (ii) the policy used to select such outer products. We experimentally show that significant improvements in computational complexity as well as accuracy can indeed be obtained through Mem-AOPGD.

cs.LG↗

arXiv2021

Decentralized optimization over noisy, rate-constrained networks: Achieving consensus by communicating differences

In decentralized optimization, multiple nodes in a network collaborate to minimize the sum of their local loss functions. The information exchange between nodes required for this task, is often limited by network connectivity. We consider a setting in which communication between nodes is hindered by both (i) a finite rate-constraint on the signal transmitted by any node, and (ii) additive noise corrupting the signal received by any node. We propose a novel algorithm for this scenario: Decentralized Lazy Mirror Descent with Differential Exchanges (DLMD-DiffEx), which guarantees convergence of the local estimates to the optimal solution under the given communication constraints. A salient feature of DLMD-DiffEx is the introduction of additional proxy variables that are maintained by the nodes to account for the disagreement in their estimates due to channel noise and rate-constraints. Convergence to the optimal solution is attained by having nodes iteratively exchange these disagreement terms until consensus is achieved. In order to prevent noise accumulation during this exchange, DLMD-DiffEx relies on two sequences; one controlling the power of the transmitted signal, and the other determining the consensus rate. We provide clear insights on the design of these two sequences which highlights the interplay between consensus rate and noise amplification. We investigate the performance of DLMD-DiffEx both from a theoretical perspective as well as through numerical evaluations.

math.OC↗

arXiv2021

Straggler Mitigation through Unequal Error Protection for Distributed Approximate Matrix Multiplication

Large-scale machine learning and data mining methods routinely distribute computations across multiple agents to parallelize processing. The time required for the computations at the agents is affected by the availability of local resources and/or poor channel conditions giving rise to the "straggler problem". As a remedy to this problem, we employ Unequal Error Protection (UEP) codes to obtain an approximation of the matrix product in the distributed computation setting to provide higher protection for the blocks with higher effect on the final result. We characterize the performance of the proposed approach from a theoretical perspective by bounding the expected reconstruction error for matrices with uncorrelated entries. We also apply the proposed coding strategy to the computation of the back-propagation step in the training of a Deep Neural Network (DNN) for an image classification task in the evaluation of the gradients. Our numerical experiments show that it is indeed possible to obtain significant improvements in the overall time required to achieve the DNN training convergence by producing approximation of matrix products using UEP codes in the presence of stragglers.

cs.DC↗

arXiv2021

Stochastic Binning and Coded Demixing for Unsourced Random Access

Unsourced random access is a novel communication paradigm designed for handling a large number of uncoordinated users that sporadically transmit very short messages. Under this model, coded compressed sensing (CCS) has emerged as a low-complexity scheme that exhibits good error performance. Yet, one of the challenges faced by CCS pertains to disentangling a large number of codewords present on a single factor graph. To mitigate this issue, this article introduces a modified CCS scheme whereby active devices stochastically partition themselves into groups that utilize separate sampling matrices with low cross-coherence for message transmission. At the receiver, ideas from the field of compressed demixing are employed for support recovery, and separate factor graphs are created for message disambiguation in each cluster. This reduces the number of active users on a factor graph, which improves performance significantly in typical scenarios. Indeed, coded demixing reduces the probability of error as the number of groups increases, up to a point. Findings are supported with numerical simulations.

cs.IT↗