Abstract: In this paper, a target signal detection method using
multiple signal classification (MUSIC) algorithm is proposed. The
MUSIC algorithm is a subspace-based direction of arrival (DOA)
estimation method. The algorithm detects the DOAs of multiple
sources using the inverse of the eigenvalue-weighted eigen spectra. To
apply the algorithm to target signal detection for GSC-based
beamforming, we utilize its spectral response for the target DOA in
noisy conditions. For evaluation of the algorithm, the performance of
the proposed target signal detection method is compared with that of
the normalized cross-correlation (NCC), the fixed beamforming, and
the power ratio method. Experimental results show that the proposed
algorithm significantly outperforms the conventional ones in receiver
operating characteristics(ROC) curves.
Abstract: Two approaches for model development of a smart acoustic box are suggested in this paper: the finite element (FE) approach and the subspace identification. Both approaches result in a state-space model, which can be used for obtaining the frequency responses and for the controller design. In order to validate the developed FE model and to perform the subspace identification, an experimental set-up with the acoustic box and dSPACE system was used. Experimentally obtained frequency responses show good agreement with the frequency responses obtained from the FE model and from the identified model.
Abstract: Many real-world data sets consist of a very high dimensional feature space. Most clustering techniques use the distance or similarity between objects as a measure to build clusters. But in high dimensional spaces, distances between points become relatively uniform. In such cases, density based approaches may give better results. Subspace Clustering algorithms automatically identify lower dimensional subspaces of the higher dimensional feature space in which clusters exist. In this paper, we propose a new clustering algorithm, ISC – Intelligent Subspace Clustering, which tries to overcome three major limitations of the existing state-of-art techniques. ISC determines the input parameter such as є – distance at various levels of Subspace Clustering which helps in finding meaningful clusters. The uniform parameters approach is not suitable for different kind of databases. ISC implements dynamic and adaptive determination of Meaningful clustering parameters based on hierarchical filtering approach. Third and most important feature of ISC is the ability of incremental learning and dynamic inclusion and exclusions of subspaces which lead to better cluster formation.
Abstract: In this paper we propose a novel approach for ascertaining human identity based on fusion of profile face and gait biometric cues The identification approach based on feature learning in PCA-LDA subspace, and classification using multivariate Bayesian classifiers allows significant improvement in recognition accuracy for low resolution surveillance video scenarios. The experimental evaluation of the proposed identification scheme on a publicly available database [2] showed that the fusion of face and gait cues in joint PCA-LDA space turns out to be a powerful method for capturing the inherent multimodality in walking gait patterns, and at the same time discriminating the person identity..
Abstract: In illumination variant face recognition, existing
methods extracting face albedo as light normalized image may lead to
loss of extensive facial details, with light template discarded. To
improve that, a novel approach for realistic facial texture
reconstruction by combining original image and albedo image is
proposed. First, light subspaces of different identities are established
from the given reference face images; then by projecting the original
and albedo image into each light subspace respectively, texture
reference images with corresponding lighting are reconstructed and
two texture subspaces are formed. According to the projections in
texture subspaces, facial texture with normal light can be synthesized.
Due to the combination of original image, facial details can be
preserved with face albedo. In addition, image partition is applied to
improve the synthesization performance. Experiments on Yale B and
CMUPIE databases demonstrate that this algorithm outperforms the
others both in image representation and in face recognition.
Abstract: This paper presents a supervised clustering algorithm,
namely Grid-Based Supervised Clustering (GBSC), which is able to
identify clusters of any shapes and sizes without presuming any
canonical form for data distribution. The GBSC needs no prespecified
number of clusters, is insensitive to the order of the input
data objects, and is capable of handling outliers. Built on the
combination of grid-based clustering and density-based clustering,
under the assistance of the downward closure property of density
used in bottom-up subspace clustering, the GBSC can notably reduce
its search space to avoid the memory confinement situation during its
execution. On two-dimension synthetic datasets, the GBSC can
identify clusters with different shapes and sizes correctly. The GBSC
also outperforms other five supervised clustering algorithms when
the experiments are performed on some UCI datasets.
Abstract: In the present work, we propose a new method for
solving the matrix equation AXB=F . The new method can
be considered as a generalized form of the well-known global full
orthogonalization method (Gl-FOM) for solving multiple linear
systems. Hence, the method will be called extended Gl-FOM (EGl-
FOM). For implementing EGl-FOM, generalized forms of block
Krylov subspace and global Arnoldi process are presented. Finally,
some numerical experiments are given to illustrate the efficiency of
our new method.
Abstract: In this paper, we present an algorithm for computing a
Schur factorization of a real nonsymmetric matrix with ordered diagonal
blocks such that upper left blocks contains the largest magnitude
eigenvalues. Especially in case of multiple eigenvalues, when matrix
is non diagonalizable, we construct an invariant subspaces with few
additional tricks which are heuristic and numerical results shows the
stability and accuracy of the algorithm.
Abstract: This article is devoted to the numerical solution of
large-scale quadratic eigenvalue problems. Such problems arise in
a wide variety of applications, such as the dynamic analysis of
structural mechanical systems, acoustic systems, fluid mechanics,
and signal processing. We first introduce a generalized second-order
Krylov subspace based on a pair of square matrices and two initial
vectors and present a generalized second-order Arnoldi process for
constructing an orthonormal basis of the generalized second-order
Krylov subspace. Then, by using the projection technique and the
refined projection technique, we propose a restarted generalized
second-order Arnoldi method and a restarted refined generalized
second-order Arnoldi method for computing some eigenpairs of largescale
quadratic eigenvalue problems. Some theoretical results are also
presented. Some numerical examples are presented to illustrate the
effectiveness of the proposed methods.
Abstract: Let T and S be a subspace of Cn and Cm, respectively.
Then for A ∈ Cm×n satisfied AT ⊕ S = Cm, the generalized
inverse A(2)
T,S is given by A(2)
T,S = (PS⊥APT )†. In this paper, a
finite formulae is presented to compute generalized inverse A(2)
T,S
under the concept of restricted inner product, which defined as <
A,B >T,S=< PS⊥APT,B > for the A,B ∈ Cm×n. By this
iterative method, when taken the initial matrix X0 = PTA∗PS⊥, the
generalized inverse A(2)
T,S can be obtained within at most mn iteration
steps in absence of roundoff errors. Finally given numerical example
is shown that the iterative formulae is quite efficient.
Abstract: Biclustering aims at identifying several biclusters that
reveal potential local patterns from a microarray matrix. A bicluster is
a sub-matrix of the microarray consisting of only a subset of genes
co-regulates in a subset of conditions. In this study, we extend the
motif of subspace clustering to present a K-biclusters clustering (KBC)
algorithm for the microarray biclustering issue. Besides minimizing
the dissimilarities between genes and bicluster centers within all
biclusters, the objective function of the KBC algorithm additionally
takes into account how to minimize the residues within all biclusters
based on the mean square residue model. In addition, the objective
function also maximizes the entropy of conditions to stimulate more
conditions to contribute the identification of biclusters. The KBC
algorithm adopts the K-means type clustering process to efficiently
make the partition of K biclusters be optimized. A set of experiments
on a practical microarray dataset are demonstrated to show the
performance of the proposed KBC algorithm.
Abstract: In the present work, we propose a new projection method for solving the matrix equation AXB = F. For implementing our new method, generalized forms of block Krylov subspace and global Arnoldi process are presented. The new method can be considered as an extended form of the well-known global generalized minimum residual (Gl-GMRES) method for solving multiple linear systems and it will be called as the extended Gl-GMRES (EGl- GMRES). Some new theoretical results have been established for proposed method by employing Schur complement. Finally, some numerical results are given to illustrate the efficiency of our new method.
Abstract: In this work, I present a review on Sparse Distributed
Memory for Small Cues (SDMSCue), a variant of Sparse Distributed
Memory (SDM) that is capable of handling small cues. I then conduct
and show some cognitive experiments on SDMSCue to test its
cognitive soundness compared to SDM. Small cues refer to input
cues that are presented to memory for reading associations; but have
many missing parts or fields from them. The original SDM failed to
handle such a problem. SDMSCue handles and overcomes this
pitfall. The main idea in SDMSCue; is the repeated projection of the
semantic space on smaller subspaces; that are selected based on the
input cue length and pattern. This process allows for Read/Write
operations using an input cue that is missing a large portion.
SDMSCue is augmented with the use of genetic algorithms for
memory allocation and initialization. I claim that SDM functionality
is a subset of SDMSCue functionality.
Abstract: In this paper we study some numerical methods to solve a model one-dimensional convection–diffusion equation. The semi-discretisation of the space variable results into a system of ordinary differential equations and the solution of the latter involves the evaluation of a matrix exponent. Since the calculation of this term is computationally expensive, we study some methods based on Krylov subspace and on Restrictive Taylor series approximation respectively. We also consider the Chebyshev Pseudospectral collocation method to do the spatial discretisation and we present the numerical solution obtained by these methods.
Abstract: Eigenvector methods are gaining increasing acceptance in the area of spectrum estimation. This paper presents a successful attempt at testing and evaluating the performance of two of the most popular types of subspace techniques in determining the parameters of multiexponential signals with real decay constants buried in noise. In particular, MUSIC (Multiple Signal Classification) and minimum-norm techniques are examined. It is shown that these methods perform almost equally well on multiexponential signals with MUSIC displaying better defined peaks.
Abstract: Sparse representation which can represent high dimensional
data effectively has been successfully used in computer vision
and pattern recognition problems. However, it doesn-t consider the
label information of data samples. To overcome this limitation,
we develop a novel dimensionality reduction algorithm namely
dscriminatively regularized sparse subspace learning(DR-SSL) in this
paper. The proposed DR-SSL algorithm can not only make use of
the sparse representation to model the data, but also can effective
employ the label information to guide the procedure of dimensionality
reduction. In addition,the presented algorithm can effectively deal
with the out-of-sample problem.The experiments on gene-expression
data sets show that the proposed algorithm is an effective tool for
dimensionality reduction and gene-expression data classification.
Abstract: In this paper, we propose novel algorithmic models
based on information fusion and feature transformation in crossmodal
subspace for different types of residue features extracted from
several intra-frame and inter-frame pixel sub-blocks in video
sequences for detecting digital video tampering or forgery. An
evaluation of proposed residue features – the noise residue features
and the quantization features, their transformation in cross-modal
subspace, and their multimodal fusion, for emulated copy-move
tamper scenario shows a significant improvement in tamper detection
accuracy as compared to single mode features without transformation
in cross-modal subspace.
Abstract: In this paper, an estimation accuracy of multiple moving
talker tracking using a microphone array is improved. The tracking
can be achieved by the adaptive method in which two algorithms are integrated, namely, the PAST (Projection Approximation Subspace
Tracking) algorithm and the IPLS (Interior Point Least Square) algorithm. When either talker begins to speak again after a silent
period, an appropriate feasible region for an evaluation function of
the IPLS algorithm might not be set. Then, the tracking fails due to the incorrect updating. Therefore, if an increment of the number of
active talkers is detected, the feasible region must be reset. Then, a low cost realization is required for the high speed tracking and a high
accuracy realization is desired for the precise tracking. In this paper,
the directions roughly estimated using the delayed-sum-array method
are used for the resetting. Several results of experiments performed in
an actual room environment show the effectiveness of the proposed method.
Abstract: Identity verification of authentic persons by their multiview faces is a real valued problem in machine vision. Multiview faces are having difficulties due to non-linear representation in the feature space. This paper illustrates the usability of the generalization of LDA in the form of canonical covariate for face recognition to multiview faces. In the proposed work, the Gabor filter bank is used to extract facial features that characterized by spatial frequency, spatial locality and orientation. Gabor face representation captures substantial amount of variations of the face instances that often occurs due to illumination, pose and facial expression changes. Convolution of Gabor filter bank to face images of rotated profile views produce Gabor faces with high dimensional features vectors. Canonical covariate is then used to Gabor faces to reduce the high dimensional feature spaces into low dimensional subspaces. Finally, support vector machines are trained with canonical sub-spaces that contain reduced set of features and perform recognition task. The proposed system is evaluated with UMIST face database. The experiment results demonstrate the efficiency and robustness of the proposed system with high recognition rates.
Abstract: The IDR(s) method based on an extended IDR theorem was proposed by Sonneveld and van Gijzen. The original IDR(s) method has excellent property compared with the conventional iterative methods in terms of efficiency and small amount of memory. IDR(s) method, however, has unexpected property that relative residual 2-norm stagnates at the level of less than 10-12. In this paper, an effective strategy for stagnation detection, stagnation avoidance using adaptively information of parameter s and improvement of convergence rate itself of IDR(s) method are proposed in order to gain high accuracy of the approximated solution of IDR(s) method. Through numerical experiments, effectiveness of adaptive tuning IDR(s) method is verified and demonstrated.