Shuxue Ding

Blind Spectral Unmixing Based on Sparse Nonnegative Matrix Factorization

Nonnegative matrix factorization (NMF) is a widely used method for blind spectral unmixing (SU), which aims at obtaining the endmembers and corresponding fractional abundances, knowing only the collected mixing spectral data. It is noted that the abundance may be sparse (i.e., the endmembers may be with sparse distributions) and sparse NMF tends to lead to a unique result, so it is intuitive and meaningful to constrain NMF with sparseness for solving SU. However, due to the abundance sum-to-one constraint in SU, the traditional sparseness measured by L0/L1-norm is not an effective constraint any more. A novel measure (termed as S-measure) of sparseness using higher order norms of the signal vector is proposed in this paper. It features the physical significance. By using the S-measure constraint (SMC), a gradient-based sparse NMF algorithm (termed as NMF-SMC) is proposed for solving the SU problem, where the learning rate is adaptively selected, and the endmembers and abundances are simultaneously estimated. In the proposed NMF-SMC, there is no pure index assumption and no need to know the exact sparseness degree of the abundance in prior. Yet, it does not require the preprocessing of dimension reduction in which some useful information may be lost. Experiments based on synthetic mixtures and real-world images collected by AVIRIS and HYDICE sensors are performed to evaluate the validity of the proposed method.

Convolutive Blind Source Separation in the Frequency Domain Based on Sparse Representation

Convolutive blind source separation (CBSS) that exploits the sparsity of source signals in the frequency domain is addressed in this paper. We assume the sources follow complex Laplacian-like distribution for complex random variable, in which the real part and imaginary part of complex-valued source signals are not necessarily independent. Based on the maximum a posteriori (MAP) criterion, we propose a novel natural gradient method for complex sparse representation. Moreover, a new CBSS method is further developed based on complex sparse representation. The developed CBSS algorithm works in the frequency domain. Here, we assume that the source signals are sufficiently sparse in the frequency domain. If the sources are sufficiently sparse in the frequency domain and the filter length of mixing channels is relatively small and can be estimated, we can even achieve underdetermined CBSS. We illustrate the validity and performance of the proposed learning algorithm by several simulation examples.

Nonnegative Blind Source Separation by Sparse Component Analysis Based on Determinant Measure

The problem of nonnegative blind source separation (NBSS) is addressed in this paper, where both the sources and the mixing matrix are nonnegative. Because many real-world signals are sparse, we deal with NBSS by sparse component analysis. First, a determinant-based sparseness measure, named D-measure, is introduced to gauge the temporal and spatial sparseness of signals. Based on this measure, a new NBSS model is derived, and an iterative sparseness maximization (ISM) approach is proposed to solve this model. In the ISM approach, the NBSS problem can be cast into row-to-row optimizations with respect to the unmixing matrix, and then the quadratic programming (QP) technique is used to optimize each row. Furthermore, we analyze the source identifiability and the computational complexity of the proposed ISM-QP method. The new method requires relatively weak conditions on the sources and the mixing matrix, has high computational efficiency, and is easy to implement. Simulation results demonstrate the effectiveness of our method.

A mobile phone-based wearable vital signs monitoring system

Design and implementation of a multiple vital signs monitoring system based upon mobile telephony and Internet infrastructure for e-health are described. The system hierarchy comprises three layers for sensing, communication, and management. The core of the sensing layer is a wearable sensor unit, including a cordless sensor device and a sensor-wear garment, suitable for vital signs real-time monitoring without discomfort and constraint during daily activities. The communication layer performs bi-directional data/command exchange via either wired or wireless means to bridge between the sensor layer and management layer. The management layer conducts comprehensive data analysis and evidence-based health management. This article describes the architecture design considerations and systemic implementation to meet various practical needs and provide scalable solutions not only for home healthcare but also other applications driven by vital signs. Three applications platformed on this architecture are explained.

Statistical mechanical properties of the q-oscillator system

The q-deformed thermofield dynamics is constructed and in terms of this dynamics some statistical mechanical properties of a system of q-harmonic oscillators are discussed.

Environmental sound recognition by multilayered neural networks

Environmental sound recognition is an important function of robotic audition. Although HMM- or TDNN-based methods can also be used for environmental sound recognition, unlike speech recognition, it is not possible to create a perfect database covering all kinds of environmental sounds. Environmental sound recognition depends more on the robot computer system task. From this point of view, the methods for environmental sound recognition must also be task-dependent and be evaluated based on accuracy, speed and simplicity. In this research, we tried to use a multilayered perceptron NN system for environmental sound recognition. The input data is the one-dimensional combination of the instantaneous spectrum at the power peak and the power pattern in time domain. The spectrum of environmental sounds do not change as remarkedly as that of speech of voice, so the combination of power and frequency pattern will retain the major features of environmental sounds but with drastically reduced data. Two experiments were conducted using an original database and a database created by the RWCP. The recognition rate for 45 environmental sound data sets was about 92%. The new method is fast and simple compared to the HMM-based methods, and suitable for an on-board system of a robot for home use, e.g. a security monitoring robot or a home-helper robot.

Diagnose the mild cognitive impairment by constructing Bayesian network with missing data

Mild Cognitive Impairment (MCI) is thought to be the prodromal phase to Alzheimers disease (AD), which is the most common form of dementia and leads to irreversible neurogenerative damage of the brain. In order to further improve the diagnostic quality of the MCI, we developed a MCI expert system to address MCIs prediction and inference question, consequently, assist the diagnosis of doctor. In this system, we mainly deal with following problems: (1) Estimate missing data in the experiment by utilizing mutual information and Newton interpolation. (2) Make certain the prior feature ordering in constructing Bayesian network. (3) Construct the Bayesian network (We term the algorithm as MNBN). The experimental results indicate that MNBN algorithm achieved better results than some existing methods in most instances. The mean square error comes to 0.0173 in the MCI experiment. Our results shed light on the potential application in MCI diagnosis.

A fast algorithm for learning overcomplete dictionary for sparse representation based on proximal operators

We present a fast, efficient algorithm for learning an overcomplete dictionary for sparse representation of signals. The whole problem is considered as a minimization of the approximation error function with a coherence penalty for the dictionary atoms and with the sparsity regularization of the coefficient matrix. Because the problem is nonconvex and nonsmooth, this minimization problem cannot be solved efficiently by an ordinary optimization method. We propose a decomposition scheme and an alternating optimization that can turn the problem into a set of minimizations of piecewise quadratic and univariate subproblems, each of which is a single variable vector problem, of either one dictionary atom or one coefficient vector. Although the subproblems are still nonsmooth, remarkably they become much simpler so that we can find a closed-form solution by introducing a proximal operator. This leads to an efficient algorithm for sparse representation. To our knowledge, applying the proximal operator to the problem with an incoherence term and obtaining the optimal dictionary atoms in closed form with a proximal operator technique have not previously been studied. The main advantages of the proposed algorithm are that, as suggested by our analysis and simulation study, it has lower computational complexity and a higher convergence rate than state-of-the-art algorithms. In addition, for real applications, it shows good performance and significant reductions in computational time.

An Efficient Method to Determine the Diagonal Loading Factor Using the Constant Modulus Feature

The diagonal loading method is a simple and efficient method that improves the robustness of beamformers. However, determining an ideal diagonal loading factor (DLF) is not a trivial problem, one that still has not been adequately addressed. Although beamformers are widely used in radar, sonar, and many other applications, in this correspondence, we consider a beamformer used only for digital communication applications. In digital communications, user information is coded into a sequence with the constant modulus (CM) in the baseband. Obviously, if the beamformer can cancel interference sufficiently well, the outputs of the beamformer should satisfy the CM condition more precisely. Therefore, minimizing CM errors can act as a natural criterion for determining DLFs. In this correspondence, we present a cost function based on the criterion. We also propose a direct search algorithm for DLF determination by minimizing the cost function. Numerical experiments show the effectiveness of the proposed method.

A near real-time approach for convolutive blind source separation

In this paper, we propose an algorithm for real-time signal processing of convolutive blind source separation (CBSS), which is a promising technique for acoustic source separation in a realistic environment, e.g., room/office or vehicle. First, we apply an overlap-and-save (sliding windows with overlapping) strategy that is most suitable for real-time CBSS processing; this approach can also aid in solving the permutation problem. Second, we consider the issue of separating sources in the frequency domain. We introduce a modified correlation matrix of observed signals and perform CBSS by diagonalization of the matrix. Third, we propose a method that can diagonalize the modified correlation matrix by solving a so-called normal equation for CBSS. One desirable feature of our proposed algorithm is that it can solve the CBSS problem explicitly, rather than stochastically, as is done with conventional algorithms. Moreover, a real-time separation of the convolutive mixtures of sources can be performed. We designed several simulations to compare the effectiveness of our algorithm with its counterpart, the gradient-based approach. Our proposed algorithm displayed superior convergence rates relative to the gradient-based approach. We also designed an experiment for testing the efficacy of the algorithm in real-time CBSS processing aimed at separating acoustic sources in realistic environments. Within this experimental context, the convergence time of our algorithms was substantially faster than that of the gradient-based algorithms. Moreover, our algorithm converges to a much lower value of the cost function than that of the gradient-based algorithm, ensuring better performance.