In this paper, we consider the problem of scheduling an agile sensor to perform an optimal control action in the case of the multi-target tracking scenario. Our purpose is to present a random finite set (RFS) approach to the multi-target sensor management problem formulated in the Partially Observed Markov Decision Process (POMDP) framework. The reward function associated with each sensor control (action) is computed via the Expected Risk Reduction between the multi-target predicted and updated densities. The proposed algorithm is implemented via the Probability Hypothesis Density filter (PHD). Numerical studies demonstrate the performance of this particular approach to a radar beam-pointing problem where targets need to be prioritized.

In this paper (This is a combined and expanded version of previous conference papers Kadri et al., 2010, 2011c) we consider the problems of supervised classification and regression in the case where attributes and labels are functions: a data is represented by a set of functions, and the label is also a function. We focus on the use of reproducing kernel Hilbert space theory to learn from such functional data. Basic concepts and properties of kernel-based learning are extended to include the estimation of function-valued functions. In this setting, the representer theorem is restated, a set of rigorously defined infinite-dimensional operator-valued kernels that can be valuably applied when the data are functions is described, and a learning algorithm for nonlinear functional data analysis is introduced. The methodology is illustrated through speech and audio signal processing experiments.

In this paper, we focus on the challenging task of the online estimation of the state and the unknown measurement noise density in nonlinear dynamic state-space models. We are especially interested in making inference in the presence of impulsive and time-varying noise. A flexible Bayesian nonparametric noise model based on an extension of Dirichlet Process Mixtures, namely the Time Varying Dirichlet process Mixtures, is introduced. A novel method based on sequential Monte Carlo methods is proposed to perform the optimal online estimation. Simulation results demonstrate the efficiency and the robustness of this method.

In this paper, we address the problem of online state and measurement noise density estimation in nonlinear dynamic state-space-models. We are especially interested in making inference in the presence of impulsive and multimodal noise. The proposed method relies on the introduction of a flexible Bayesian nonparametric noise model based on Dirichlet Process mixtures. A novel approach based on sequential Monte Carlo methods is proposed to perform the optimal online estimation. Simulation results demonstrate the efficiency and the robustness of this approach.

The Probability Hypothesis Density (PHD) is a well-known method for single-sensor multi-target tracking problems in a Bayesian framework, but the extension to the multi-sensor case seems to remain a challenge. In this paper, an extension of Mahler's work to the multi-sensor case provides an expression of the true PHD multi-sensor data update equation. Then, based on the configuration of the sensors' fields of view (FOVs), a joint partitioning of both the sensors and the state space provides an equivalent yet more practical expression of the data update equation, allowing a more effective implementation in specific FOV configurations.

In this paper, we tackle the problem of interference mitigation in ad hoc networks. In such context, the multiple access interference (MAI) is known to be of an impulsive nature. Therefore, the conventional Gaussian assumption can not be considered to model this type of interference. Contrariwise, it can be accurately modeled by stable distributions. Here, this issue is addressed within an Orthogonal Frequency Division Multiplexing (OFDM) transmission link assuming a symmetric α-stable model for the signal distortion due to MAI. For this purpose, we propose a method for the joint estimation of the transmitted multicarrier signal and the noise parameters. Based on sequential Monte Carlo (SMC) methods, the proposed scheme allows the online estimation using a Rao-blackwellized particle filter.

The problem of acoustic-to-articulatory speech inversion continues to be a challenging research problem which significantly impacts automatic speech recognition robustness and accuracy. This paper presents a multi-task kernel based method aimed at learning Vocal Tract (VT) variables from the Mel-Frequency Cepstral Coefficients (MFCCs). Unlike usual speech inversion techniques based on individual estimation of each tract variable, the key idea here is to consider all the target variables simultaneously to take advantage of the relationships among them and then improve learning performance. The proposed method is evaluated using synthetic speech dataset and corresponding tract variables created by the TAsk Dynamics Application (TADA) model and compared to the hierarchical ε-SVR speech inversion technique.

The Dirichlet Process Mixture (DPM) models represent an attractive approach to modeling latent distributions parametrically. In DPM models the Dirichlet process (DP) is applied especially when the distribution of latent parameters is to be considered as multimodal. DPMs allow for uncertainty in the choice of parametric forms and in the number of mixing components (clusters). The parameters of a DP include the precision a and the base probability measure G0(μ, Σ). In most applications, the choice of priors and posteriors computation for the hyperparameters (α, μ, Σ) clearly influences inferences about the level of clustering in the mixture. This is the main focus of this paper. We consider the problem of density estimation of an observation noise distribution in a dynamic nonlinear model from a Bayesian nonparametric view-point. Our approach is illustrated in a real-world data analysis task dealing with the estimation of pseudorange errors in a GNSS based localization context.

This paper shows a method for the modeling of speech signal distributions based on Dirichlet process mixtures (DPM) and the estimation of noise sequences based on particle filtering. In real situations, the speech recognition rate degrades miser ably because of the effect of environmental noises, reflected waves and so on. To improve the speech recognition rate, a technique for the estimation of noise sequences is necessary. In this paper, the distribution of the clean speech is modeled using the DPM instead of the traditional model, which is a Gaussian mixture model (GMM). Speech signal sequences are generated according to the mean and covariance generated from the DPM. Then, noise signal sequences are estimated with a particle filter. The proposed method using extended Kalman filter (EKF) can improve the speech recognition rate significantly in the low SNR region. Applying unscented Kalman filter (UKF), better results can be obtained in also the high SNR.

The question tackled here is the time allocation of radars in a multitarget environment. At a given time radars can only observe a limited part of the space; it is therefore necessary to move their axis with respect to time, in order to be able to explore the overall space facing them. Such sensors are used to detect, to locate and to identify targets which are in their surrounding aerial space. In this paper we focus on the detection schema when several targets need to be detected by a set of delocalized radars. This work is based on the modelling of the radar detection performances in terms of probability of detection and on the optimization of a criterion based on detection probabilities. This optimization leads to the derivation of allocation strategies and is made for several contexts and several hypotheses about the targets locations.

Using Kalman techniques, it is possible to perform optimal estimation in linear Gaussian state-space models. We address here the case where the noise probability density functions are of unknown functional form. A flexible Bayesian nonparametric noise model based on mixture of Dirichlet processes is introduced. Efficient Markov chain Monte Carlo and sequential Monte Carlo methods are then developed to perform optimal estimation in such contexts

We consider here the case where our knowledge is partial and based on a betting density function which is n-dimensional Gaussian. The explicit formulation of the least committed basic belief density (bbd) of the multivariate Gaussian pdf is provided. Beliefs are then assigned to hyperspheres and the bbd follows a chi2distribution. An application to model based classification in the joint speed-acceleration feature space is presented

Until now, humanitarian demining has been unable to provide a solution to the landmine removal problem. Furthermore, new low-cost methods have to be developed quickly. While much progress has been made with the introduction of new sensor types, other problems have been raised by these sensors. Ground-penetrating radars (GPRs) are key sensors for landmine detection as they are capable of detecting landmines with low metal contents. GPRs deliver so-called Bscan data, which are, roughly, vertical slice images of the ground. However, due to the high dielectric permittivity contrast at the air–ground interface, a strong response is recorded at an early time by GPRs. This response is the main component of the so-called clutter noise, and it blurs the responses of landmines buried at shallow depths. The landmine detection task is therefore quite difficult, and a preprocessing step, which aims at reducing the clutter, is often needed. In this paper, a difficult case for clutter reduction, that is, when landmines and clutter responses overlap in time, is presented. A new and simple clutter removal method based on the design of a two-dimensional digital filter, which is adapted to Bscan data, is proposed. The designed filter must reduce the clutter on Bscan data significantly while protecting the landmine responses. In order to do so, a frequency analysis of a clutter geometrical model is first led. Then, the same process is applied to a geometrical model of a signal coming from a landmine. This results in building a high-pass digital filter and determining its cutoff frequencies. Finally, simulations are presented on simulated and real data, and a comparison with the classical clutter removal algorithm is made.

Ground-penetrating radars (GPRs) are very promising sensors for landmine detection as they are capable of detecting landmines with low metal contents. GPRs deliver so-called Bscan data which are, roughly, vertical slice images of the ground. However, due to the high dielectric permittivity contrast at the air-ground interface, a strong response is recorded at early time by GPRs. This response is the main component of the so-called clutter noise and it blurs the responses of landmines buried at shallow depths. The landmine detection task is therefore quite difficult. This paper proposes a new method for automated detection and localization of buried objects from Bscan records. A support vector machine algorithm for online abrupt change detection is implemented and proves to be efficient in detecting buried landmines from Bscan data. The proposed procedure performance is evaluated using simulated and real data.

In the frame of humanitarian antipersonnel mines detection, a multisensor fusion method using the Dempster-Shafer evidence theory is presented. The multisensor system consists of two sensors-a ground penetrating radar (GPR) and a metal detector (MD). For each sensor, a new features extraction method is presented. The method for the GPR is mainly based on wavelets and contours extraction. First simulations on a limited set of data show that an improvement in detection and false alarms rejection, for the GPR as a standalone sensor, could be obtained. The MD features extraction method is mainly based on contours extraction. All of these features are then fused with the GPR ones in some specific cases in order to determine a new feature. From these results, belief functions, as defined in the evidence theory, are then determined and combined thanks to the orthogonal sum. First results in terms of detection and false alarm rates are presented for a limited set of real data and a comparison is made between the two cases: with or without multisensor fusion.

In the frame of humanitarian land mine detection, a feature extraction method from ground penetrating radar (GPR) signals is presented. Based on the use of wavelets, this method should allow one to obtain a weighted decision for the GPR sensor and improvements in detection and false alarm rejection. A multisensor fusion principle with the weighted decision from the metal detector sensor is then presented.

The authors consider the management of two heterogeneous sensors in a multitarget environment. The sensors are a radar and an infrared camera. They describe the first results of their work: the modeling of sensors, the determination of a criterion which takes into account all the parameters of the problem and the optimization of the criterion leading to the derivation of management strategies of both sensors. These results prove that this approach allows the determination of management for the sensors and also allows the derivation of performance of the management methods which could be used to compare several possible management strategies.

Most studies dealing with mines detection emphasise sensors and processing linked to them. However, most people agree that one sensor cannot be the solution to the detection problems: a system including several sensors seems to be the solution. It implies that a data fusion function must be realized in this system. We propose in this paper the derivation of a fusion operator based on fuzzy logic. We also expose a sensitivity study of this operator with respect to several parameters as the depth or the geometry of the estimated shape of a buried detected object.