arXiv subjects

Stefano Rini

Publications and source records attributed to Stefano Rini.

At least 55 records · Page 3Linked to original sources

arXiv2021

Wireless Federated Learning with Limited Communication and Differential Privacy

This paper investigates the role of dimensionality reduction in efficient communication and differential privacy (DP) of the local datasets at the remote users for over-the-air computation (AirComp)-based federated learning (FL) model. More precisely, we consider the FL setting in which clients are prompted to train a machine learning model by simultaneous channel-aware and limited communications with a parameter server (PS) over a Gaussian multiple-access channel (GMAC), so that transmissions sum coherently at the PS globally aware of the channel coefficients. For this setting, an algorithm is proposed based on applying federated stochastic gradient descent (FedSGD) for training the minimum of a given loss function based on the local gradients, Johnson-Lindenstrauss (JL) random projection for reducing the dimension of the local updates, and artificial noise to further aid user's privacy. For this scheme, our results show that the local DP performance is mainly improved due to injecting noise of greater variance on each dimension while keeping the sensitivity of the projected vectors unchanged. This is while the convergence rate is slowed down compared to the case without dimensionality reduction. As the performance outweighs for the slower convergence, the trade-off between privacy and convergence is higher but is shown to lessen in high-dimensional regime yielding almost the same trade-off with much less communication cost.

cs.IT↗

arXiv2021

Comparison-limited Vector Quantization

In this paper a variation of the classic vector quantization problem is considered. In the standard formulation, a quantizer is designed to minimize the distortion between input and output when the number of reconstruction points is fixed. We consider, instead, the scenario in which the number of comparators used in quantization is fixed. More precisely, we study the case in which a vector quantizer of dimension d is comprised of k comparators, each receiving a linear combination of the inputs and producing the output value one/zero if this linear combination is above/below a certain threshold. In reconstruction, the comparators' output is mapped to a reconstruction point, chosen so as to minimize a chosen distortion measure between the quantizer input and its reconstruction. The Comparison-Limited Vector Quantization (CLVQ) problem is then defined as the problem of optimally designing the configuration of the compactors and the choice of reconstruction points so as to minimize the given distortion. In this paper, we design a numerical optimization algorithm for the CLVQ problem. This algorithm leverages combinatorial geometrical notions to describe the hyperplane arrangement induced by the configuration of the comparators. It also relies on a genetic genetic meta heuristic to improve the selection of the quantizer initialization and avoid local minima encountered during optimization. We numerically evaluate the performance of our algorithm in the case of input distributions following uniform and Gaussian i.i.d. sources to be compressed under quadratic distortion and compare it to the classic Linde-Buzo-Gray (LBG) algorithm.

cs.IT↗

arXiv2021

Straggler Mitigation through Unequal Error Protection for Distributed Matrix Multiplication

Large-scale machine learning and data mining methods routinely distribute computations across multiple agents to parallelize processing. The time required for computation at the agents is affected by the availability of local resources giving rise to the "straggler problem" in which the computation results are held back by unresponsive agents. For this problem, linear coding of the matrix sub-blocks can be used to introduce resilience toward straggling. The Parameter Server (PS) utilizes a channel code and distributes the matrices to the workers for multiplication. It then produces an approximation to the desired matrix multiplication using the results of the computations received at a given deadline. In this paper, we propose to employ Unequal Error Protection (UEP) codes to alleviate the straggler problem. The resiliency level of each sub-block is chosen according to its norm as blocks with larger norms have higher effects on the result of the matrix multiplication. We validate the effectiveness of our scheme both theoretically and through numerical evaluations. We derive a theoretical characterization of the performance of UEP using random linear codes, and compare it the case of equal error protection. 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 which we investigate the fundamental trade-off between precision and the time required for the computations.

cs.IT↗

arXiv2021

Hierarchical Causal Bandit

Causal bandit is a nascent learning model where an agent sequentially experiments in a causal network of variables, in order to identify the reward-maximizing intervention. Despite the model's wide applicability, existing analytical results are largely restricted to a parallel bandit version where all variables are mutually independent. We introduce in this work the hierarchical causal bandit model as a viable path towards understanding general causal bandits with dependent variables. The core idea is to incorporate a contextual variable that captures the interaction among all variables with direct effects. Using this hierarchical framework, we derive sharp insights on algorithmic design in causal bandits with dependent arms and obtain nearly matching regret bounds in the case of a binary context.

stat.ML↗

arXiv2021

Multi-Class Unsourced Random Access via Coded Demixing

Unsourced random access (URA) is a recently proposed communication paradigm attuned to machine-driven data transfers. In the original URA formulation, all the active devices share the same number of bits per packet. The scenario where several classes of devices transmit concurrently has so far received little attention. An initial solution to this problem takes the form of group successive interference cancellation, where codewords from a class of devices with more resources are recovered first, followed by the decoding of the remaining messages. This article introduces a joint iterative decoding approach rooted in approximate message passing. This framework has a concatenated coding structure borrowed from the single-class coded compressed sensing and admits a solution that offers performance improvement at little added computational complexity. Our findings point to new connections between multi-class URA and compressive demixing. The performance of the envisioned algorithm is validated through numerical simulations.

cs.IT↗

arXiv2021

The Rate-Distortion Risk in Estimation from Compressed Data

Consider the problem of estimating a latent signal from a lossy compressed version of the data when the compressor is agnostic to the relation between the signal and the data. This situation arises in a host of modern applications when data is transmitted or stored prior to determining the downstream inference task. Given a bitrate constraint and a distortion measure between the data and its compressed version, let us consider the joint distribution achieving Shannon's rate-distortion (RD) function. Given an estimator and a loss function associated with the downstream inference task, define the rate-distortion risk as the expected loss under the RD-achieving distribution. We provide general conditions under which the operational risk in estimating from the compressed data is asymptotically equivalent to the RD risk. The main theoretical tools to prove this equivalence are transportation-cost inequalities in conjunction with properties of compression codes achieving Shannon's RD function. Whenever such equivalence holds, a recipe for designing estimators from datasets undergoing lossy compression without specifying the actual compression technique emerges: design the estimator to minimize the RD risk. Our conditions simplified in the special cases of discrete memoryless or multivariate normal data. For these scenarios, we derive explicit expressions for the RD risk of several estimators and compare them to the optimal source coding performance associated with full knowledge of the relation between the latent signal and the data.

cs.IT↗

arXiv2020

On Sum Secure Degrees of Freedom for K-User MISO Broadcast Channel With Alternating CSIT

In this paper, the sum secure degrees of freedom (SDoF) of the $K$-user Multiple Input/Single Output (MISO) Broadcast Channel with Confidential Messages (BCCM) and alternating Channel State Information at the Transmitter (CSIT) is investigated. In the MISO BCCM, a $K$-antenna transmitter (TX) communicates toward $K$ single-antenna receivers (RXs), so that message for RX $k$ is kept secret from RX $j$ with $j<k$. For this model, we consider the scenario in which the CSI of the RXs from $2$ to $K$ is instantaneously known at the transmitter while CSI of RX $1$ is known at the transmitter (i) instantaneously for half of the time and (ii) with a unit delay for the remainder of the time. We refer to this CSIT availability as \emph{alternating} CSIT. Alternating CIST has been shown to provide synergistic gains in terms of SDoF and is thus of a viable strategy to ensure secure communication by simply relying on the CSI feedback strategy. Our main contribution is the characterization of sum SDoF for this model as $SDoF_{\rm sum}= (2K-1)/2$. Interestingly, this $SDoF_{\rm sum}$ is attained by a rather simple achievability in which the TX uses artificial noise to prevent the decoding of the message of the unintended receivers at RX $1$. For simplicity first, the proof for the case $K=3$ is discussed in detail and after that, we have presented the results for any number of RXs.

cs.IT↗

arXiv2020

An Exploration of the Heterogeneous Unsourced MAC

The unsourced MAC model was originally introduced to study the communication scenario in which a number of devices with low-complexity and low-energy wish to upload their respective messages to a base station. In the original problem formulation, all devices communicate using the same information rate. This may be very inefficient in certain wireless situations with varied channel conditions, power budgets, and payload requirements at the devices. This paper extends the original problem setting so as to allow for such variability. More specifically, we consider the scenario in which devices are clustered into two classes, possibly with different SNR levels or distinct payload requirements. In the cluster with higher power,devices transmit using a two-layer superposition modulation. In the cluster with lower energy, users transmit with the same base constellation as in the high power cluster. Within each layer, devices employ the same codebook. At the receiver, signal groupings are recovered using Approximate Message Passing(AMP), and proceeding from the high to the low power levels using successive interference cancellation (SIC). This layered architecture is implemented using Coded Compressed Sensing(CCS) within every grouping. An outer tree code is employed to stitch fragments together across times and layers, as needed.This pragmatic approach to heterogeneous CCS is validated numerically and design guidelines are identified.

cs.IT↗

arXiv2020

An efficient label-free analyte detection algorithm for time-resolved spectroscopy

Time-resolved spectral techniques play an important analysis tool in many contexts, from physical chemistry to biomedicine. Customarily, the label-free detection of analytes is manually performed by experts through the aid of classic dimensionality-reduction methods, such as Principal Component Analysis (PCA) and Non-negative Matrix Factorization (NMF). This fundamental reliance on expert analysis for unknown analyte detection severely hinders the applicability and the throughput of these such techniques. For this reason, in this paper, we formulate this detection problem as an unsupervised learning problem and propose a novel machine learning algorithm for label-free analyte detection. To show the effectiveness of the proposed solution, we consider the problem of detecting the amino-acids in Liquid Chromatography coupled with Raman spectroscopy (LC-Raman).

cs.LG↗

arXiv2020

Fairness-Oriented User Association in HetNets Using Bargaining Game Theory

In this paper, the user association and resource allocation problem is investigated for a two-tier HetNet consisting of one macro Base Station (BS) and a number of pico BSs. The effectiveness of user association to BSs is evaluated in terms of fairness and load distribution. In particular, the problem of determining a fair user association is formulated as a bargaining game so that for the Nash Bargaining Solution (NBS) abiding the fairness axioms provides an optimal and fair user association. The NBS also yields in a Pareto optimal solution and leads to a proportional fair solution in the proposed HetNet model. Additionally, we introduce a novel algorithmic solution in which a new Coalition Generation Algorithm (CGA), called SINR-based CGA, is considered in order to simplify the coalition generation phase. Our simulation results show the efficiency of the proposed user association scheme in terms of fairness and load distribution among BSs and users. In particular, we compare the performance of the proposed solution with that of the throughput-oriented scheme in terms of the max-sum-rate scheme and show that the proposed solution yields comparable average data rates and overall sum rate.

cs.IT↗

arXiv2020

Efficient Federated Learning over Multiple Access Channel with Differential Privacy Constraints

In this paper, the problem of federated learning (FL) through digital communication between clients and a parameter server (PS) over a multiple access channel (MAC), also subject to differential privacy (DP) constraints, is studied. More precisely, we consider the setting in which clients in a centralized network are prompted to train a machine learning model using their local datasets. The information exchange between the clients and the PS takes places over a MAC channel and must also preserve the DP of the local datasets. Accordingly, the objective of the clients is to minimize the training loss subject to (i) rate constraints for reliable communication over the MAC and (ii) DP constraint over the local datasets. For this optimization scenario, we proposed a novel consensus scheme in which digital distributed stochastic gradient descent (D-DSGD) is performed by each client. To preserve DP, a digital artificial noise is also added by the users to the locally quantized gradients. The performance of the scheme is evaluated in terms of the convergence rate and DP level for a given MAC capacity. The performance is optimized over the choice of the quantization levels and the artificial noise parameters. Numerical evaluations are presented to validate the performance of the proposed scheme.

cs.LG↗

arXiv2020

The Information & Mutual Information Ratio for Counting Image Features and Their Matches

Feature extraction and description is an important topic of computer vision, as it is the starting point of a number of tasks such as image reconstruction, stitching, registration, and recognition among many others. In this paper, two new image features are proposed: the Information Ratio (IR) and the Mutual Information Ratio (MIR). The IR is a feature of a single image, while the MIR describes features common across two or more images.We begin by introducing the IR and the MIR and motivate these features in an information theoretical context as the ratio of the self-information of an intensity level over the information contained over the pixels of the same intensity. Notably, the relationship of the IR and MIR with the image entropy and mutual information, classic information measures, are discussed. Finally, the effectiveness of these features is tested through feature extraction over INRIA Copydays datasets and feature matching over the Oxfords Affine Covariant Regions. These numerical evaluations validate the relevance of the IR and MIR in practical computer vision tasks

cs.CV↗

arXiv2020

Decentralized SGD with Over-the-Air Computation

We study the performance of decentralized stochastic gradient descent (DSGD) in a wireless network, where the nodes collaboratively optimize an objective function using their local datasets. Unlike the conventional setting, where the nodes communicate over error-free orthogonal communication links, we assume that transmissions are prone to additive noise and interference.We first consider a point-to-point (P2P) transmission strategy, termed the OAC-P2P scheme, in which the node pairs are scheduled in an orthogonal fashion to minimize interference. Since in the DSGD framework, each node requires a linear combination of the neighboring models at the consensus step, we then propose the OAC-MAC scheme, which utilizes the signal superposition property of the wireless medium to achieve over-the-air computation (OAC). For both schemes, we cast the scheduling problem as a graph coloring problem. We numerically evaluate the performance of these two schemes for the MNIST image classification task under various network conditions. We show that the OAC-MAC scheme attains better convergence performance with a fewer communication rounds.

eess.SP↗

arXiv2020

On the Capacity of the Oversampled Wiener Phase Noise Channel

In this paper, the capacity of the oversampled Wiener phase noise (OWPN) channel is investigated. The OWPN channel is a discrete-time point-to-point channel with a multi-sample receiver in which the channel output is affected by both additive and multiplicative noise. The additive noise is a white standard Gaussian process while the multiplicative noise is a Wiener phase noise process. This channel generalizes a number of channel models previously studied in the literature which investigate the effects of phase noise on the channel capacity, such as the Wiener phase noise channel and the non-coherent channel. We derive upper and inner bounds to the capacity of OWPN channel: (i) an upper bound is derived through the I-MMSE relationship by bounding the Fisher information when estimating a phase noise sample given the past channel outputs and phase noise realizations, then (ii) two inner bounds are shown: one relying on coherent combining of the oversampled channel outputs and one relying on non-coherent combining of the samples. After capacity, we study generalized degrees of freedom (GDoF) of the OWPN channel for the case in which the oversampling factor grows with the average transmit power $P$ as $P$? and the frequency noise variance as $P^α$?. Using our new capacity bounds, we derive the GDoF region in three regimes: regime (i) in which the GDoF region equals that of the classic additive white Gaussian noise (for $β\leq 1$), one (ii) in which GDoF region reduces to that of the non-coherent channel (for $β\geq \min \{α,1\}$) and, finally, one in which partially-coherent combining of the over-samples is asymptotically optimal (for $2 α-1\leq β\leq 1$). Overall, our results are the first to identify the regimes in which different oversampling strategies are asymptotically optimal.

cs.IT↗

arXiv2019

Comparison-limited Vector Quantization

A variation of the classic vector quantization problem is considered, in which the analog-to-digital (A2D) conversion is not constrained by the cardinality of the output but rather by the number of comparators available for quantization. More specifically, we consider the scenario in which a vector quantizer of dimension d is comprised of k comparators, each receiving a linear combination of the inputs and producing zero/one when this signal is above/below a threshold. Given a distribution of the inputs and a distortion criterion, the value of the linear combinations and thresholds are to be configured so as to minimize the distortion between the quantizer input and its reconstruction. This vector quantizer architecture naturally arises in many A2D conversion scenarios in which the quantizer's cost and energy consumption are severely restricted. For this novel vector quantizer architecture, we propose an algorithm to determine the optimal configuration and provide the first performance evaluation for the case of uniform and Gaussian sources.

cs.IT↗

arXiv2019

LDPC Coded Multiuser Shaping for the Gaussian Multiple Access Channel

The joint design of input constellation and low-density parity-check (LDPC) codes to approach the symmetric capacity of the two-user Gaussian multiple access channel is studied. More specifically, multilevel coding is employed at each user to construct a high-order input constellation and the constellations of the users are jointly designed so as to maximize the multiuser shaping gain. At the receiver, each layer of the multilevel coding is jointly decoded among users, while successive cancellation is employed across layers. The LDPC code employed by each user in each layer is designed using EXIT charts to support joint decoding among users for the prescribed per-layer rate and SNR. Numerical simulations are provided to validate the proposed constellation and LDPC code designs.

cs.IT↗

arXiv2018

Distributed Convex Optimization With Limited Communications

In this paper, a distributed convex optimization algorithm, termed \emph{distributed coordinate dual averaging} (DCDA) algorithm, is proposed. The DCDA algorithm addresses the scenario of a large distributed optimization problem with limited communication among nodes in the network. Currently known distributed subgradient methods, such as the distributed dual averaging or the distributed alternating direction method of multipliers algorithms, assume that nodes can exchange messages of large cardinality. Such network communication capabilities are not valid in many scenarios of practical relevance. In the DCDA algorithm, on the other hand, communication of each coordinate of the optimization variable is restricted over time. For the proposed algorithm, we bound the rate of convergence under different communication protocols and network architectures. We also consider the extensions to the case of imperfect gradient knowledge and the case in which transmitted messages are corrupted by additive noise or are quantized. Relevant numerical simulations are also provided.

cs.DC↗

arXiv2018

On MIMO Channel Capacity with Output Quantization Constraints

The capacity of a Multiple-Input Multiple-Output (MIMO) channel in which the antenna outputs are processed by an analog linear combining network and quantized by a set of threshold quantizers is studied. The linear combining weights and quantization thresholds are selected from a set of possible configurations as a function of the channel matrix. The possible configurations of the combining network model specific analog receiver architectures, such as single antenna selection, sign quantization of the antenna outputs or linear processing of the outputs. An interesting connection between the capacity of this channel and a constrained sphere packing problem in which unit spheres are packed in a hyperplane arrangement is shown. From a high-level perspective, this follows from the fact that each threshold quantizer can be viewed as a hyperplane partitioning the transmitter signal space. Accordingly, the output of the set of quantizers corresponds to the possible regions induced by the hyperplane arrangement corresponding to the channel realization and receiver configuration. This connection provides a number of important insights into the design of quantization architectures for MIMO receivers; for instance, it shows that for a given number of quantizers, choosing configurations which induce a larger number of partitions can lead to higher rates.

cs.IT↗