Ammar Mesloub

Logistic regression is a fundamental and widely used statistical method for modeling binary outcomes based on covariates. However, the presence of missing data, particularly in settings involving hybrid covariates (a mix of discrete and continuous variables), poses significant challenges. In this paper, we propose a novel Expectation-Maximization based algorithm tailored for parameter estimation in logistic regression models with missing hybrid covariates. The proposed method is specifically designed to handle these complexities, delivering efficient parameter estimates. Through comprehensive simulations and real-world application, we demonstrate that our approach consistently outperforms traditional methods, achieving superior accuracy and reliability.

In the context of condition-based maintenance of rotating machines in manufacturing systems, the early diagnosis of possible faults related to rolling elements of the bearing is mainly based on techniques from artificial intelligence, namely, Machine Learning (ML) and Deep Learning (DL). Approaches based on using Deep Learning methods have been the most coveted in recent years. Among a variety of models, the type of architecture known as Long-Short-Term Memory (LSTM) of Recurrent Neural Network (RNN) has both the ability to capture long-term dependencies and to adapt to sequential data modeling. It is therefore able to work on data without any preprocessing. This paper studies using four types of LSTM networks to diagnose bearing faults in a classification approach. It aims to intervene on both the input parameters and the network architecture, to achieve high performance. The proposed method is carried out in two different ways. In the first case, the data inputs are raw frames of vibration signals. However, in the second case, the network inputs are pre-computed time-frequency features. The results clearly showed that LSTMs are more accurate with the latter.

The Joint Singular Value Decomposition (JSVD) aims to find a pair of unitary matrices allowing a simultaneous diagonalization of a set of matrices. Based on Givens rotations, this paper proposes two approaches dealing with the JSVD problem in the complex domain. A new criterion is defined to set all diagonal entries to real and non-negative values. An indirect method based on complex to real transformation and a direct method exploiting justified steps of the first one are developed. Numerical simulations show the performance achieved by our algorithm in terms of JSVD quality and convergence rate.

Like the Joint Diagonalization of a unique set of matrices, Generalized Joint Diagonalization is an algebraic problem encountered in different applications such as data fusion and blind source separation. This paper proposes a new generalized unitary joint diagonalization approach based on the Jacobi iterative scheme using Givens rotations and simplified criterion, introducing three approximations. These approximations allowed us to reach a simultaneous estimation of different parameters. The first appears in the simplified criterion composed of entries doubly affected by the Givens rotations. The second approximation is in the Givens parameter using a small amplitude angle. The last approximation resides in keeping only the first order of transformed entries. Numerical experiments, including examples of joint blind audio source separation, are provided. The results show the effectiveness of the developed algorithm as compared to existing ones. The simultaneous estimation of different Givens rotations improves the algorithm's computation complexity and convergence rate

In this paper, a new approximate joint diagonalization problem is formulated for the blind separation of possibly dependent sources, modelled as widely linear autoregressive moving average (WL-ARMA) signals. The latter modelling gives rise to two sets of matrix parameters, sharing each a specific diagonal structure transform. A new method is proposed to achieve the desired source separation using a novel extended joint eigen value decomposition (JEVD) of both sets of matrices. The separation method proceeds in two steps, the first one identifies the WL-ARMA parameters, while the second step identifies the separation matrix through the extended JEVD of two particular matrix sets. We have developed a new Jacobi-like JEVD algorithm based on Shear and Givens rotations. This algorithm, referred to as Extended Optimal Phase Joint Eigen Value Decomposition, realizes the simultaneous diagonalization of two sets of matrices with different structures and outperforms the existing algorithms in difficult scenarios. Simulation results are provided to illustrate the effectiveness of the proposed algorithm.

Respiration rate monitoring using ultra-wideband (UWB) radar is preferred because it provides contactless measurement without restricting the person’s privacy. This study considers a novel non-contact-based solution using a single-input multiple-output (SIMO) UWB impulse radar. In the proposed system, the collected radar data are converted to several narrow-band signals using the generalized Goertzel algorithm (GGA), which are used as the input of the designed phased arrays for position estimation. In this context, we introduce the incoherent signal subspace methods (ISSM) for the direction of arrivals (DOAs) and distance evaluation. Meanwhile, a beam focusing approach is used to determine each individual and estimate their breathing rate automatically based on a linearly constrained minimum variance (LCMV) beamformer. The experimental results prove that the proposed algorithm can achieve high estimation accuracy in a variety of test environments, with an error of 2%, 5%, and 2% for DOA, distance, and respiration rate, respectively.

Impulse Radio Ultra-Wide-Band (IR-UWB) radar is a promising technology for contact-less sensing applications, such as Vital Sign (VS) detection of human body. This paper presents a new approach for Respiration Rate (RR) estimation of a stationary person. The particularity of this approach is that does not require any pre-processing techniques. It is based on the estimation of the Second Maximum Eigenvalue (SME) using Singular Value Decomposition (SVD). To extract the VS signal, a packet of eigenvalues is acquired using a moving matrix window over the slow time direction of the radar data. Then applying a Fast Fourier Transform (FFT), the respiration rate can be easily estimated. Experiments are performed using a P440 (IR-UWB) radar. The results are encouraging and show that the proposed method effectively detects the VS with high accuracy even in low Signal to Noise Ratio (SNR) condition, such as in through the wall human detection scenario.

This letter deals with a hybrid joint diagonalization problem considering both Hermitian and transpose congruence. Such problem can be encountered in certain noncircular signal analysis applications including blind source separation. We introduce new Jacobi-like algorithms using Givens or a combination of Givens and hyperbolic rotations. These algorithms are compared with state-of-the-art methods and their performance gain, especially in the high dimensional case, is assessed through simulation experiments including examples related to blind separation of noncircular sources.