Ricardo Suyama

• PhD in Electrical Engineering
• Federal University of ABC

Blind Source Separation (BSS) is a well-known problem in signal processing and still receives attention from the scientific community, given its applicability in different areas. This work presents a theoretical background overview of the Kalman Filter formulation and its applicability to the BSS problem as a parameter estimator in two different ap...

In this book, written in Portuguese, we discuss what ill-posed problems are and how the regularization method is used to solve them. In the form of questions and answers, we reflect on the origins and future of regularization, relating the similarities and differences of its meaning in different areas, including inverse problems, statistics, machin...

This study focuses on Intelligent Fault Diagnosis (IFD) in rotating machinery utilizing a single microphone and a data-driven methodology, effectively diagnosing 42 classes of fault types and severities. The research leverages sound data from the imbalanced MaFaulDa dataset, aiming to strike a balance between high performance and low resource consu...

Massive multiple-input multiple-output (MMIMO) is essential to modern wireless communication systems, like 5G and 6G, but it is vulnerable to active eavesdropping attacks. One type of such attack is the pilot contamination attack (PCA), where a malicious user copies pilot signals from an authentic user during uplink, intentionally interfering with...

Time-Difference Electrical Impedance Tomography (TDEIT) is an imaging technique to visualize resistivity changes over time in a region of interest. Regularization is necessary because TDEIT is an ill-posed problem. In this work, we use Regularization by Denoising (RED) with four different denoisers to reconstruct brain images in a simplified 2D hea...

Electrical Impedance Tomography (EIT) is an imaging modality that allows the visualization of internal resistivities of a region of interest from electrical measurements external to the same region. In this work, we reconstruct 3D static images using two regularization terms, an anatomical atlas with \(\ell _1\)-norm and a total variation (TV) term...

Electrical impedance tomography (EIT) is a medical imaging modality that has the potential to benefit diagnosing, monitoring, and understanding several pathological conditions. However, some regions of the body, such as the brain, are more challenging to reconstruct, demanding improvements before the technique can be used in clinical practice. In t...

Objetivo: A Inteligência Artificial se mostra promissora como apoio à decisão no câncer de mama, porém, a interpretabilidade dos algoritmos como os de caixa preta pode contribuir na adoção na prática clínica. Esse estudo apresenta a explicabilidade em Modelos Preditivos de Aprendizado de Máquina no Câncer de Mama. Métodos: Avaliou-se duas abordagen...

Graph theory has been widely and efficiently applied to characterize brain functioning, ranging from the diagnosis of pathologies (e.g. depression, Alzheimer, schizophrenia, etc.) to investigations of cognitive processes in neuroscience. Recently, the use of graph-based strategies through functional connectivity (FC) analysis has shown to be an int...

Correntropy has been successfully used for signal processing in a large variety of applications. Usually, correntropy estimator uses a Gaussian kernel, which leads to a measure that takes into account all even order statistical moments of the underlying signal. In this paper we analyze correntropy implemented with the Epanechnikov kernel, which is...

In several Wireless Sensor Networks applications, the sensitive nature of collected information makes the security of data transmission a major concern. Different cryptography techniques have been considered to provide reliable communication between nodes. However, the secure distribution of encryption keys is still a challenging task. In this work...

Different approaches have been investigated to implement effective Brain-Computer Interfaces (BCI), translating brain activation patterns into commands to external devices. BCI exploring Steady-State Visually Evoked Potentials usually achieve relatively high accuracy, when considering 2-3 second sample windows, but the performance degrades for smal...

Este trabalho apresenta os estudos associados à realização do sensoriamento de forças aplicadas em uma prótese de mão robótica. O intuito é que o usuário possa controlar a força aplicada a determinados objetos sem que estes sofram deformações plásticas mas que estejam seguros para que não haja o deslizamento do objeto. Para isto, foi empregado o pr...

A correntropia vem sendo utilizada no processamento de sinais em uma ampla gama de aplicações. Normalmente, a correntropia utiliza um kernel gaussiano, o que faz com que a medida leve em conta todos os momentos estatísticos de ordem par dos sinais envolvidos. Neste artigo, analisamos a correntropia implementada com o kernel Epanechnikov. Visto que...

Os diferentes métodos de aprendizado de máquina geram modelos que possuem peculiaridades em sua natureza, sejam elas na forma como atuam sobre os dados, na divisão das regiões de classificação ou em relação à existência de fatores intrínsecos, como a aleatoriedade. Neste trabalho, é apresentada uma análise comparativa de desempenho entre as redes M...

Identifying musical instruments in a song is a challenging task due the complex nature of the problem. However, great efforts have been dedicated to this task aiming at solving some issues such as, music tagging or identifying unknown instruments used in music tracks and in genre detection. In order to understand the problem and propose a possible...

The work assessed seven classical classifiers and two beamforming algorithms for detecting surveillance sound events. The tests included the use of AWGN with-10 dB to 30 dB SNR. Data Augmentation was also employed to improve algorithms' performance. The results showed that the combination of SVM and Delay-and-Sum (DaS) scored the best accuracy (up...

Objective. Adapted from the concept of channel capacity, the information transfer rate (ITR) has been widely used to evaluate the performance of a brain–computer interface (BCI). However, its traditional formula considers the model of a discrete memoryless channel in which the transition matrix presents very particular symmetries. As an alternative...

Due to the growing demand for improving surveillance capabilities in smart cities, systems need to be developed to provide better monitoring capabilities to competent authorities, agencies responsible for strategic resource management, and emergency call centers. This work assumes that, as a complementary monitoring solution, the use of a system ca...

Speech enhancement is a crucial task for several applications. Among the most explored techniques are the Wiener filter and the LogMMSE, but approaches exploring deep learning adapted to this task, such as SEGAN, have presented relevant results. This study compared the performance of the mentioned techniques in 85 noise conditions regarding quality...

In this paper, we present some preliminary results using the Speech Enhancement Generative Adversarial Network (SEGAN) for the attenuation of the ego-noise in the speech source localization problem embedded in unmanned aerial vehicles (UAV). This task is of great interest in UAV search and rescue scenarios. The primary motivation of using the SEGAN...

This work aims to investigate different techniques for the classification of atypical sound events, considering the accuracy and classification time, in order to find which of these techniques are most advantageous for surveillance (security) scenarios, minimizing the possibility of relevant events to go unnoticed, specially due to human failures....

The recognition of audio signals is a process that has been used in several activities with different purposes, such as: Therapeutics (phonoaudiological), Legal (voice identification), elimination of noise, among others. This paper presents a method of recognizing musical instruments timbre based on the kNN Pattern Classification process.

This work investigates an alternative approach to the problem of blind equalization. The approach is based on complexity measures and is inspired by preceding successful application of the same framework to the problem of blind source separation. We draw the relationship between algorithmic complexity , a measure for randomness within the area of a...

The wireless physical layer security can assist traditional encryption mechanisms by generating the keys, for instance. However, some techniques allow hiding information directly in the physical layer, such as performing precoding at the legitimate user using information about the channel, such as the received signal strength or the phase response....

This work aims to investigate different techniques for the classification of atypical sound events, considering the accuracy and classification time, in order to find which of these techniques are most advantageous for surveillance scenarios, minimizing the possibility of relevant events to go unnoticed. LDA, QDA, Decision Tree and KNN techniques w...

This work has a twofold aim: (a) to analyze an alternative approach for computing the conditional Lyapunov exponent (λcmax) aiming to evaluate the synchronization stability between nonlinear oscillators without solving the classical variational equations for the synchronization error dynamical system. In this first framework, an analytic reference...

Minimum Entropy Deconvolution (MED) is a sparse blind deconvolution method that searches for a deconvolution filter that leads to the most sparse output, assuming that the desired signal is originally sparse. The present work establishes sufficient conditions for the blind deconvolution of sparse images. Then, based on a measure of sparsity given b...

Brain-computer interfaces (BCIs) outline alternative communication channels with the objective to directly map brain signals onto control signals for external devices. In this context, this paper presents the performance of the canonical correlation analysis (CCA) as a feature extraction method and also as a preprocessing strategy for a BCI system...

The use of Independent Component Analysis (ICA) for pre-processing or selecting features of EEG signals derived from motor imagery tasks has been the benchmark option in the literature. In this work the combined use of ICA and feature selection techniques was capable of enhancing task imagery responses in Brain-Computer Interface (BCI) systems, whi...

The Izhikevich neuronal model has been adopted as an important benchmark in theoretical neuroscience, and exhibits one of the best relations between biological plau-sibility and computational cost. The present work aims to analyze the synchronization between unidirectional coupled Izhikevich neurons by means of the conditional Lyapunov exponent eva...

The sensorimotor areas of the brain respond in a similar way to an executed movement and to a motor imagery task. Hence, it becomes possible to classify different brain states only with the electroencephalography (EEG) signal arising from the imagination e.g. of the movement of either the left or right hand. The use of oscillations found in the sen...

Brain-Computer Interface (BCI) systems outline alternative communication channels that map brain signals directly onto an external application, which can be extremely useful to improve the quality of life for severely disable people. One of the most employed paradigms for designing BCI systems is based on steady-state visually evoked potentials (SS...

With the increasing number of machine learning problems that are out of the linear and Gaussian paradigm, information theoretic learning (ITL) rises as a research field that proposes a modeling method with a wealthier statistical treatment of the adaptation criterion. In the first part of this tutorial, we introduce the main concepts of ITL and a k...

This is the second part of the introductory tutorial about information theoretic learning, which, after the theoretical foundations presented in Part I, now discusses the concepts of correntropy, a new similarity measure derived from the quadratic entropy, and presents example problems where the ITL framework can be successfully applied: dynamic mo...

The characterization of nonlinear dynamical systems and their attractors in terms of invariant measures, basins of attractions and the structure of their vector fields usually outlines a task strongly related to the underlying computational cost. In this work, the practical aspects related to the use of parallel computing - specially the use of Gra...

This work, which is intended to be a celebration to the 35 years of the constant modulus criterion and to the impact it had in the development of our research group, presents both tutorial elements and a discussion of published and unpublished results that, hopefully, will generate new reflections and perspectives on this important topic.

In this work, we investigate the use of monotonic neural networks as compensating functions in the context of source separation of post-nonlinear (PNL) mixtures. We first provide a numerical example that illustrates the importance of having bijective nonlinear compensating functions in PNL models. Then, we propose a separation framework in which a...

In this paper, we propose a method for blind compensation of a memoryless nonlinear distortion. We assume as prior information that the desired signal admits a sparse representation in a transformed domain that should be known in advance. Then, given that a nonlinear distortion tends to generate signals that are less sparse than the desired one, ou...

The separation of an underdetermined audio mixture can be performed through sparse component analysis (SCA) that relies however on the strong hypothesis that source signals are sparse in some domain. To overcome this difficulty in the case where the original sources are available before the mixing process, the informed source separation (ISS) embed...

This paper presents a simple and, to a certain extent, surprising result for Source Separation in an underdetermined scenario: without loss of generality, under the restriction that all sources have unit power, the sum of the residual mean-squared errors (MMSE) obtained after the estimation of all the sources is given by the difference between the...

Recentemente, concebeu-se uma formulação polinomial do critério CM, a qual permitiu a obtenção de um limitante inferior para a função custo CM e de soluções analíticas baseadas na relaxação de restrições [1]. Neste trabalho, apresentaremos uma visão geométrica dessa formulação.

The feature extraction stage is one of the main tasks underlying pattern recognition, and, is particularly important for designing Brain-Computer Interfaces (BCIs), i.e. structures capable of mapping brain signals in commands for external devices. Within one of the most used BCIs paradigms, that based on Steady State Visual Evoked Potentials (SSVEP...

In this work, an extended polynomial formulation of the constant modulus (CM) criterion under quadratic constraints is presented. Based on the method of Lagrange Multipliers, this `Volterra-CM formulation' brings very relevant information about the structure of the null-gradient CM solutions in the equalizer parameter space, including a conjecture...

In this work, we present a novel bioinspired framework for performing ICA over finite (Galois) fields of prime order P. The proposal is based on a state-of-the-art immune-inspired algorithm, the cob-aiNet[C], which is employed to solve a combinatorial optimization problem — associated with a minimal entropy configuration — adopting a Michigan-like...

As an aperiodic bounded dynamics in a deterministic system, chaos has been found in various fields of science and technology for the past 50 years. In the last two decades, we have witnessed a rapid development in chaos control (synchronization) theory and its application, that since the seminal works referred to as the OGY control and PC synchroni...

This paper addresses the problem of direction-of-arrival (DOA) parameter estimation in array processing when the signals are inherently discrete, which is the case mainly in the digital communication context. Based on the particular structure of the signal space in the data model, a maximum likelihood-based approach is introduced. The strategy cons...

This work presents a method for denoising chaotic time series when the structure of the underlying dynamics is known, albeit not the associated initial conditions and parameters. The strategy relies on finding the initial conditions and free parameters that minimize deviations – in the mean-squared error sense – from the noisy observations, thus pr...

The recently proposed lower bound of the classical blind CM criterion was shown to be able to work as an excellent blind equalizability index, which is a practical performance assessment metric in the context of inverse problems. However, there still remains the need of complementary studies aiming to cover a wider range of circumstances. Based on...

From the classical blind CM criterion, we derive a MSE-based polynomial formulation that lead to two contribu-tions: a lower bound of the CM criterion, which works as an equalizability index, and an initialization heuristic for the CMA. The results indicate the validity of the index as an analytical tool and as a practical performance assessment me...

The blind deconvolution of signals composed of statistically dependent samples is an important practical problem whose understanding still requires the clarification of many theoretical points. In this work, we present an analysis of this problem that includes two well-established methods -the canonical constant modulus algorithm (CMA) and a corren...

Brain-computer interfaces (BCIs) provide new alternative channels for human communication, being of particular importance as assistive instruments for severe disable people. Such undoubtedly relevance led to the development of several BCIs working paradigms as the ones outlined by motor imagery or by steady-state visual evoked potentials (SSVEP). I...

The present work aims to apply a recently proposed method for estimating Lyapunov exponents to characterize-with the aid of the metric entropy and the fractal dimension-the degree of information and the topological structure associated with multiscroll attractors. In particular, the employed methodology offers the possibility of obtaining the whole...

This article introduces a blind equalization methodology based on a cascade of two-tap finite impulse response filters globally optimal with respect to the constant modulus (CM) criterion. It is shown that a formulation of the CM cost function in terms of a mean squared error (MSE) metric and of Volterra-like constraints allow the optimization proc...

The development of brain-computer interfaces (BCIs) for disabled patients is currently a growing field of research. Most BCI systems are based on electroencephalography (EEG) signals, and within this group, systems using motor imagery (MI) are amongst the most flexible. However, for stroke patients, the motor areas of the brain are not always avail...

Classically, adaptive equalization algorithms are analyzed in terms of two possible steady state behaviors: convergence to a fixed point and divergence to infinity. This twofold scenario suits well the modus operandi of linear supervised algorithms, but can be rather restrictive when unsupervised methods are considered, as their intrinsic use of hi...

Recently, many chaos-based communication systems have been proposed. They can present the many interesting properties of spread spectrum modulations. Besides, they can represent a low-cost increase in security. However, their major drawback is to have a Bit Error Rate (BER) general performance worse than their conventional counterparts. In this pap...

In this paper,we address the problem of blind compensation of nonlinear distortions. Our approach relies on the assumption that the input signal is bandlimited. We then make use of the classical result that the output of a nonlinearity has a wider spectrum than the one of the input signal. However,differently from previous works,our approach does n...

The problem of optimal prediction was developed predominantly under the aegis of the assumption that all signal samples and system parameters should be real or complex numbers. In this work, we provide an extension of this problem to the framework of finite (Galois) fields through use of the classical and elegant framework defined by stochastic mod...

In the context of temporally correlated sources, blind deconvolution can be performed using the well-known constant modulus algorithm and, most recently, also using approaches based on information theory, like correntropy. However, given the different nature of both classes, a certain degree of discrepant behavior is expected when they are compared...

In this work, we present a new interpretation of recurrent separation strategies in terms of local inversion based on iterative methods for solving nonlinear equation systems. From this interpretation, we firstly obtain a fresh perspective on the important methods proposed by Hérault and Jutten and by Hosseini and Deville for solving, respectively,...

In this work, we address the problem of compensating a nonlinear memoryless system in a blind fashion, i.e., without considering a set of training points. Our proposal works with the assumption that the input signal admits a sparse representation in a transformed domain that should be known in advance. By assuming that the nonlinear distortion func...

Unsupervised signal processing has been an exciting theme of research for at least three decades. It finds the potential application in practically all fields where well-established techniques of digital signal processing have been employed, including telecommunications; speech and audio processing; image, radar, and sonar; and biomedical signals....

Multiscroll attractors for low dimensional dynamical systems exhibit an interesting commitment between regularity and phase space exploration. Usually, its most common form is obtained from variants of the classical Chua's circuit model, which implies in analyzing non-smooth piecewise linear state equations, something that can be a laborious task b...

In this work, we present an original implementation of a circuit to perform the analog simulation of the FitzHugh-Nagumo neuron model. The proposed circuit is tested for different stimulation patterns and provides results very similar to those derived from a more widespread digital computation approach, with a significantly better performance in te...

Chaotic synchronization in master-slave networks has been extensively studied in the last years, with a relevant impact in application domains like communication systems and the modeling of neuronal and other biomedical signals and systems. Many recent papers have shown that chaotic synchronization is easily lost when there is additive noise in the...

The problem of optimal linear filtering and prediction has been so far typically formulated and studied in the context of real-or complex-valued signals. In this article, we provide an extension of this problem to the framework of finite (Galois) fields. Simulation results encompassing supervised and unsupervised prediction-based equalization are p...

This work presents a new method to calculate the Lyapunov spectrum of dynamical systems based on the time evolution of initially small disturbed copies (“clones”) of the motion equations. In this approach, it is not necessary to construct the tangent space associated with the time evolution of linearized versions of motion equations, being the Lyap...

This work has a twofold aim: to present a numerical analysis of the Hodgkin–Huxley model in a nonsmooth excitation scenario — which is both challenging and theoretically relevant — and to use the established framework as a basis for testing a method to search for specific oscillating patterns in dynamical systems. The analysis is founded on classic...

The problem of independent component analysis (ICA) was firstly formulated and studied in the context of real-valued signals and mixing models, but, recently, an extension of this original formulation was proposed to deal with the problem within the framework of finite fields. In this work, we propose a strategy to deal with ICA over these fields t...

We investigate the application of cost functions based on the ℓ0-norm to the problem of blind source extraction (BSE). We show that if the sources have different levels of sparsity, then the minimization of the ℓ0-norm leads to the extraction of the sparsest component even when the sources are statistically dependent. We also study the conditions g...

This work aims to present a new method to perform blind extraction of chaotic deterministic sources mixed with stochastic signals. This technique employs the recurrence quantification analysis (RQA), a tool commonly used to study dynamical systems, to obtain the separating system that recovers the deterministic source. The method is applied to inve...