Yikui ZHAI Junying GAN Jinwen LI Junying ZENG Ying XU
Due to security demand of society development, real-time face recognition has been receiving more and more attention nowadays. In this paper, a real-time face recognition system via Local Binary Pattern (LBP) plus Improved Biomimetic Pattern Recognition (BPR) has been proposed. This system comprises three main steps: real-time color face detection process, feature extraction process and recognition process. Firstly, a color face detector is proposed to detect face with eye alignment and simultaneous performance; while in feature extraction step, LBP method is adopted to eliminate the negative effect of the light heterogeneity. Finally, an improved BPR method with Selective Sampling construction is applied to the recognition system. Experiments on our established database named WYU Database, PUT Database and AR Database show that this real-time face recognition system can work with high efficiency and has achieved comparable performance with the state-of-the-art systems.
Myung-Ho PARK Ki-Gon NAM Jin Seok KIM Dae Hyun YUM Pil Joong LEE
With the increased deployment of wireless sensor networks (WSNs) in location-based services, the need for accurate localization of sensor nodes is gaining importance. Sensor nodes in a WSN localize themselves with the help of anchors that know their own positions. Some anchors may be malicious and provide incorrect information to the sensor nodes. In this case, accurate localization of a sensor node may be severely affected. In this study, we propose a secure and lightweight localization method. In the proposed method, uncertainties in the estimated distance between the anchors and a sensor node are taken into account to improve localization accuracy. That is, we minimize the weighted summation of the residual squares. Simulation results show that our method is very effective for accurate localization of sensor nodes. The proposed method can accurately localize a sensor node in the presence of malicious anchors and it is computationally efficient.
Ildefons MAGRANS DE ABRIL Masashi SUGIYAMA
This letter presents the ideas and methods of the winning solution* for the Kaggle Algorithmic Trading Challenge. This analysis challenge took place between 11th November 2011 and 8th January 2012, and 264 competitors submitted solutions. The objective of this competition was to develop empirical predictive models to explain stock market prices following a liquidity shock. The winning system builds upon the optimal composition of several models and a feature extraction and selection strategy. We used Random Forest as a modeling technique to train all sub-models as a function of an optimal feature set. The modeling approach can cope with highly complex data having low Maximal Information Coefficients between the dependent variable and the feature set and provides a feature ranking metric which we used in our feature selection algorithm.
Zhengliang LV Shiyuan YANG Hong WANG Linda MILOR
Process variation causes significant fluctuations in the timing performance of analog circuits, which causes a fraction of circuits to fail specifications. By testing the delay-performance, we can recognize the failed circuits during production testing. In this paper, we have proposed a low overhead and process tolerant delay evaluation circuit for built-in self test (BIST) function for analog differential circuits. This circuit contains a delay generation cell, an input differential signal generation cell, a delay matching cell, a sample-hold circuit, and a comparator. This circuit was implemented with 0.18 µm CMOS process. Simulation results over process variation, devices mismatch and layout parasitics, but without silicon measurement, show that the accuracy in delay detection is within 5 ps. A case study was done over a feed-forward equalizer (FFE). A typical use of this circuit is testing the delay of various FIR (Finite Impulse Response) filters.
Peng GONG Ping LI Duk Kyung KIM
In this letter, unlike the previous work in [2], the optimal power allocation in a non-orthogonal, amplify-and-forward (AF) relay-assisted transmission is investigated in the uplink. Here, the inter-user-interference among the signals from MTs and relays exists due to non-zero interference suppression factor (ISF), i.e., finite spreading factor. In this letter, we show that the optimal solution to achieve a 'max-min fairness' among mobile terminals can be alternatively obtained by solving its inverse problem. The impact of various ISFs as well as the Jain's fairness is investigated in comparison with the equal power allocation.
Pei-Chun LIN Junzo WATADA Berlin WU
The objective of our research is to build a statistical test that can evaluate different risks of a portfolio selection model with fuzzy data. The central points and radiuses of fuzzy numbers are used to determine the portfolio selection model, and we statistically evaluate the best return by a fuzzy statistical test. Empirical studies are presented to illustrate the risk evaluation of the portfolio selection model with interval values. We conclude that the fuzzy statistical test enables us to evaluate a stable expected return and low risk investment with different choices for k, which indicates the risk level. The results of numerical examples show that our method is suitable for short-term investments.
Masashi KIYOMI Toshiki SAITOH Ryuhei UEHARA
PREIMAGE CONSTRUCTION problem by Kratsch and Hemaspaandra naturally arose from the famous graph reconstruction conjecture. It deals with the algorithmic aspects of the conjecture. We present an O(n8) time algorithm for PREIMAGE CONSTRUCTION on permutation graphs and an O(n4(n+m)) time algorithm for PREIMAGE CONSTRUCTION on distance-hereditary graphs, where n is the number of graphs in the input, and m is the number of edges in a preimage. Since each graph of the input has n-1 vertices and O(n2) edges, the input size is O(n3) (, or O(nm)). There are polynomial time isomorphism algorithms for permutation graphs and distance-hereditary graphs. However the number of permutation (distance-hereditary) graphs obtained by adding a vertex to a permutation (distance-hereditary) graph is generally exponentially large. Thus exhaustive checking of these graphs does not achieve any polynomial time algorithm. Therefore reducing the number of preimage candidates is the key point.
Dong-Ju KIM Sang-Heon LEE Myoung-Kyu SHON
This paper proposes a novel face recognition approach using a centralized gradient pattern image and image covariance-based facial feature extraction algorithms, i.e. a two-dimensional principal component analysis and an alternative two-dimensional principal component analysis. The centralized gradient pattern image is obtained by AND operation of a modified center-symmetric local binary pattern image and a modified local directional pattern image, and it is then utilized as input image for the facial feature extraction based on image covariance. To verify the proposed face recognition method, the performance evaluation was carried out using various recognition algorithms on the Yale B, the extended Yale B and the CMU-PIE illumination databases. From the experimental results, the proposed method showed the best recognition accuracy compared to different approaches, and we confirmed that the proposed approach is robust to illumination variation.
Mariko HIROKAWA Yoshimitsu KUROKI
PCA-L1 (principal component analysis based on L1-norm maximization) is an approximate solution of L1-PCA (PCA based on the L1-norm), and has robustness against outliers compared with traditional PCA. However, the more dimensions the feature space has, the more calculation time PCA-L1 consumes. This paper focuses on an initialization procedure of PCA-L1 algorithm, and proposes a fast method of PCA-L1 using Gram-Schmidt orthogonalization. Experimental results on face recognition show that the proposed method works faster than conventional PCA-L1 without decrease of recognition accuracy.
Hirotoshi HONMA Kodai ABE Yoko NAKAJIMA Shigeru MASUYAMA
Let Gs=(Vs, Es) be a simple connected graph. A vertex v ∈ Vs is an articulation vertex if deletion of v and its incident edges from Gs disconnects the graph into at least two connected components. Finding all articulation vertices of a given graph is called the articulation vertex problem. A vertex u ∈ Vs is called a hinge vertex if there exist any two vertices x and y in Gs whose distance increase when u is removed. Finding all hinge vertices of a given graph is called the hinge vertex problem. These problems can be applied to improve the stability and robustness of communication network systems. In this paper, we propose linear time algorithms for the articulation vertex problem and the hinge vertex problem of circular permutation graphs.
Jia-Rui LIU Shi-Ze GUO Zhe-Ming LU Fa-Xin YU Hui LI
In complex network analysis, there are various measures to characterize the centrality of each node within a graph, which determines the relative importance of each node. The more centrality a node has in a network, the more significance it has in the spread of infection. As one of the important extensions to shortest-path based betweenness centrality, the flow betweenness centrality is defined as the degree to which each node contributes to the sum of maximum flows between all pairs of nodes. One of the drawbacks of the flow betweenness centrality is that its time complexity is somewhat high. This Letter proposes an approximate method to calculate the flow betweenness centrality and provides experimental results as evidence.
Yong-Ho LEE Hyukchan KWON Jin-Mok KIM Kiwoong KIM Kwon-Kyu YU In-Seon KIM Chan-Seok KANG Seong-Joo LEE Seong-Min HWANG Yong-Ki PARK
As sensitive magnetic sensors, magnetometers based on superconducting quantum interference devices can be used for the detection of weak magnetic fields. These signals can be generated by diverse origins, for example, brain electric activity, myocardium electric activity, and nuclear precession of hydrogen protons. In addition, weak current induced in the low-temperature detectors, for example, transition-edge sensors can be detected using SQUIDs. And, change of magnetic flux quantum generated in a superconducting ring can be detected by SQUID, which can be used for realization of mechanical force. Thus, SQUIDs are key elements in precision metrology. In Korea, development of low-temperature SQUIDs based on Nb-Josephson junctions was started in late 1980s, and Nb-based SQUIDs have been used mainly for biomagnetic measurements; magnetocardiography and magnetoencephalography. High-Tc SQUIDs, being developed in mid 1990s, were used for magnetocardiography and nondestructive evaluation. Recently, SQUID-based low-field nuclear magnetic resonance technology is under development. In this paper, we review the past progress and recent activity of SQUID applications in Korea, with focus on biomagnetic measurements.
Masayuki HIRATA Toshiki YOSHIMINE
Magnetoencephalography (MEG) measures very weak neuromagnetic signals using SQUID sensors. Standard MEG analyses include averaged waveforms, isofield maps and equivalent current dipoles. Beamforming MEG analyses provide us with frequency-dependent spatiotemporal information about the cerebral oscillatory changes related to not only somatosensory processing but also language processing. Language dominance is able to be evaluated using laterality of power attenuation in the low γ band in the frontal area. Neuromagnetic signals of the unilateral upper movements are able to be decoded using a support vector machine.
This paper presents a basic investigation of the power imbalance problem with regard to maximum ratio combining (MRC) array antennas for digital TV broadcast reception. First, the relationship between the decrease in the diversity gain and reduction in the received power was investigated using two-element and four-element dipole array antennas by means of a Monte Carlo simulation. The relationship between the decrease in the diversity gain and the number of branches imposed to reduce the received power was also investigated. Then, a simple method of predicting the reduction in the diversity gain under imbalanced power conditions is given using the simulation results. The objective is to determine a criterion associated with the gain reduction that allows us to achieve the required system performance. Finally, the proposed method is confirmed by analysis using a model representing a typical portable digital broadcasting TV set held with both hands that simulates the power imbalance condition.
Yuelin MA Yasushi YAMAO Yoshihiko AKAIWA
Compensation for the nonlinear systems represented by polynomials involves polynomial inverse. In this paper, a new algorithm is proposed that gives the baseband polynomial inverse with a limited order. The algorithm employs orthogonal basis that is predetermined from the distribution of input signal and finds the coefficients of the inverse polynomial to minimize the mean square error. Compared with the well established p-th order inverse method, the proposed method can suppress the distortions better including higher order distortions. It is also extended to obtain memory polynomial inverse through a feedback-configured structure. Both numerical simulations and experimental results demonstrate that the proposed algorithm can provide good performance for compensating the nonlinear systems represented by baseband polynomials.
De-Chun SUN Zu-Jun LIU Ke-Chu YI
In precoded TDD MIMO systems, precoding is done based on the downlink CSI, which can be predicted according to the outdated uplink CSI. This letter proposes a double-scale channel prediction scheme where frame-scale Kalman filters and pilot-symbol-scale AR predictors jointly predict the needed downlink CSI.
This paper presents a power analysis that applies to elliptic curves over generalized Mersenne prime field Fp. This prime field enables efficient modular reductions which influence the computational performance of an elliptic curve cryptosystem. The general modular reductions stochastically calculate extra operations. Some studies showed the possibility of power analysis attacks to scalar multiplication with a unified code by using the statistical information of extra operations. In this paper, we present the statistical experiment and possibility of attacks, and propose the more sensitive attack and the countermeasure without performance impact.
Lin SHAN Sonia AISSA Hidekazu MURATA Susumu YOSHIDA Liang ZHAO
The important issue of an adaptive scheduling scheme is to maximize throughput while providing fair services to all users, especially under strict quality of service requirements. To achieve this goal, we consider the problem of multiuser scheduling under a given fairness constraint. A novel Adaptive Fairness and Throughput Control (AFTC) approach is proposed to maximize the network throughput while attaining a given min-max fairness index. Simulation results reveal that comparing to straightforward methods, the proposed AFTC approach can achieve the desired fairness while maximizing the throughput with short convergence time, and is stable in dynamic scenarios. The trade-off between fairness and throughput can be accurately controlled by adjusting the scheduler's parameters.
Xianglei XING Sidan DU Hua JIANG
We extend the Nonparametric Discriminant Analysis (NDA) algorithm to a semi-supervised dimensionality reduction technique, called Semi-supervised Nonparametric Discriminant Analysis (SNDA). SNDA preserves the inherent advantages of NDA, that is, relaxing the Gaussian assumption required for the traditional LDA-based methods. SNDA takes advantage of both the discriminating power provided by the NDA method and the locality-preserving power provided by the manifold learning. Specifically, the labeled data points are used to maximize the separability between different classes and both the labeled and unlabeled data points are used to build a graph incorporating neighborhood information of the data set. Experiments on synthetic as well as real datasets demonstrate the effectiveness of the proposed approach.
Xuefeng BAI Tiejun ZHANG Chuanjun WANG Ahmed A. ABD EL-LATIF Xiamu NIU
Player detection is an important part in sports video analysis. Over the past few years, several learning based detection methods using various supervised two-class techniques have been presented. Although satisfactory results can be obtained, a lot of manual labor is needed to construct the training set. To overcome this drawback, this letter proposes a player detection method based on one-class SVM (OCSVM) using automatically generated training data. The proposed method is evaluated using several video clips captured from World Cup 2010, and experimental results show that our approach achieves a high detection rate while keeping the training set construction's cost low.