Showing papers by "Jeng-Shyang Pan published in 2013"

Fuzzy Forecasting Based on Two-Factors Second-Order Fuzzy-Trend Logical Relationship Groups and Particle Swarm Optimization Techniques

Abstract: In this paper, we present a new method for fuzzy forecasting based on two-factors second-order fuzzy-trend logical relationship groups and particle swarm optimization (PSO) techniques. First, we fuzzify the historical training data of the main factor and the secondary factor, respectively, to form two-factors second-order fuzzy logical relationships. Then, we group the two-factors second-order fuzzy logical relationships into two-factors second-order fuzzy-trend logical relationship groups. Then, we obtain the optimal weighting vector for each fuzzy-trend logical relationship group by using PSO techniques to perform the forecasting. We also apply the proposed method to forecast the Taiwan Stock Exchange Capitalization Weighted Stock Index and the NTD/USD exchange rates. The experimental results show that the proposed method gets better forecasting performance than the existing methods.

Fuzzy Rules Interpolation for Sparse Fuzzy Rule-Based Systems Based on Interval Type-2 Gaussian Fuzzy Sets and Genetic Algorithms

Abstract: In this paper, we present a new method for fuzzy rules interpolation for sparse fuzzy rule-based systems based on interval type-2 Gaussian fuzzy sets and genetic algorithms. First, we present a method to deal with the interpolation of fuzzy rules based on interval type-2 Gaussian fuzzy sets. We also prove that the proposed method guarantees to produce normal interval type-2 Gaussian fuzzy sets. Then, we present a method to learn optimal interval type-2 Gaussian fuzzy sets for sparse fuzzy rule-based systems based on genetic algorithms. We also apply the proposed fuzzy rules interpolation method and the proposed learning method to deal with multivariate regression problems and time series prediction problems. The experimental results show that the proposed fuzzy rules interpolation method using the optimally learned interval type-2 Gaussian fuzzy sets gets higher average accuracy rates than the existing methods.

An improvement to the nearest neighbor classifier and f ace recognition experiments

Abstract: The conventional nearest neighbor classier (NNC) directly exploits the dis- tances between the test sample and training samples to perform classication. NNC independently evaluates the distance between the test sample and a training sample. In this paper, we propose to use the classication procedure of sparse representation to im- prove NNC. The proposed method has the following basic idea: the training samples are not uncorrelated and the \distance" between the test sample and a training sample should not be independently calculated and should take into account the relationship between dif- ferent training samples. The proposed methodrst uses a linear combination of all the training samples to represent the test sample and then exploits modied \distance" to classify the test sample. The method obtains the coefficients of the linear combination by solving a linear system. The method then calculates the distance between the test sample and the result of multiplying each training sample by the corresponding coefficient and assumes that the test sample is from the same class as the training sample that has the minimum distance. The method elaborately modies NNC and considers the relationship between different training samples, so it is able to produce a higher classication accu- racy. A large number of face recognition experiments on three face image databases show that the maximum difference between the accuracies of the proposed method and NNC is greater than 10%.

Fault Node Recovery Algorithm for a Wireless Sensor Network

Abstract: This paper proposes a fault node recovery algorithm to enhance the lifetime of a wireless sensor network when some of the sensor nodes shut down. The algorithm is based on the grade diffusion algorithm combined with the genetic algorithm. The algorithm can result in fewer replacements of sensor nodes and more reused routing paths. In our simulation, the proposed algorithm increases the number of active nodes up to 8.7 times, reduces the rate of data loss by approximately 98.8%, and reduces the rate of energy consumption by approximately 31.1%.

Low-Latency Digit-Serial and Digit-Parallel Systolic Multipliers for Large Binary Extension Fields

Abstract: For cryptographic algorithms, such as elliptic curve digital signature algorithm (ECDSA) and pairing algorithm, the crypto-processors are required to perform large number of additions and multiplications over finite fields of large orders To have a balanced trade-off between space complexity and time complexity, in this paper, novel digit-serial and digit-parallel systolic structures are presented for computing multiplication over GF(2m) Based on novel decomposition algorithm, we have derived an efficient digit-serial systolic architecture, which involves latency of O(√{m/d}) clock cycles, while the existing digit-serial systolic multipliers involve at least O(m/d) latency for digit-size d The proposed digit-serial design could be used for AESP-based fields with the same digit-size as the case of trinomial-based fields with a small increase in area We have also proposed digit-parallel systolic architecture employing n-term Karatsuba-like method, where the latency can be reduced from O(√{m/d}) to O(√{m/nd}) This feature would be a major advantage for implementing multiplication for the fields of large orders From synthesis results, it is shown that the proposed architectures have significantly lower time complexity, lower area-delay product, and higher bit-throughput than the existing digit-serial multipliers

A Scalable Transitive Human-Verifiable Authentication Protocol for Mobile Devices

Abstract: The man-in-the-middle (MITM) attack is the major threat for handheld devices to agree on a session key in which they do not share any prior secret in advance, even if these devices are physically located in the same place. Apart from insecurely typing passwords into handheld devices or comparing long hexadecimal keys displayed on the devices' screens, many other human-verifiable protocols have been proposed in the literature to solve the problem. Unfortunately, most of these schemes are unscalable to more users. Even when there are only three entities attempting to agree on a session key, these protocols need to be rerun three times. In this paper, we present a bipartite and a tripartite authentication protocol using a temporary confidential channel. Besides, we further extend the system into a transitive authentication protocol that allows multiple handheld devices to establish a conference key securely and efficiently. In addition, we provide a formal proof to our protocol to demonstrate our scheme is indeed secure. We also implement the prototype of the system on a mobile phone with satisfying performance.

Kernel Learning Algorithms for Face Recognition

Abstract: Kernel Learning Algorithms for Face Recognition covers the framework of kernel based face recognition. This book discusses the advanced kernel learning algorithms and its application on face recognition. This book also focuses on the theoretical deviation, the system framework and experiments involving kernel based face recognition. Included within are algorithms of kernel based face recognition, and also the feasibility of the kernel based face recognition method. This book provides researchers in pattern recognition and machine learning area with advanced face recognition methods and its newest applications.

Face recognition based on fusion of multi-resolution Gabor features

Abstract: In this paper, we propose a Gabor-based face recognition method. This method fuses multi-resolution Gabor features of face images at the matching score level. The first implementation scheme of this method directly takes the sum of the matching scores of multi-resolution Gabor features of face images as the final matching score. The second implementation scheme first codes the phase of the Gabor feature and then uses a weighted matching score level fusion algorithm to fuse the magnitude and phase of the Gabor feature. A number of experimental results show that the proposed method has a good performance and outperforms conventional Gabor-based face recognition methods that equally treat all the Gabor features and directly fuse them at the feature level. The experimental result also illustrates that in face recognition, the low-resolution representation of the phase of the Gabor feature such as the code of the phase is more discriminative than the phase itself. The codes of our method will be available at http://www.yongxu.org/lunwen.html.

Grade Diffusion Algorithm

Abstract: In this paper, a grade diffusion algorithm is proposed to solve the sensor node’s transmission problem and the sensor node’s loading problem in wireless sensor networks by to arrange the sensor node’s routing. In addition to them, the sensor node also can save some backup nodes to reduce the energy consumption for the re-looking routing by our proposed algorithm in case the sensor node’s routing is broken. In the simulation, the grade diffusion algorithm can save 28.66% energy and increase 76.39% lift time than the tradition algorithms for sensor node. Moreover, our proposed algorithm has the less data package transmission loss and the hop count than the tradition algorithms in our simulate setting. Hence, in addition to balance the sensor node’s loading and reduce the energy consumption, our algorithm can send the data package to destination node quickly and correctly.

A Reduce Identical Event Transmission Algorithm for Wireless Sensor Networks

Abstract: This paper proposed a Reduce Identical Event Transmission Algorithm (RIET). The algorithm can decide that which sensor nodes could send the event to sink node when sensor nodes sense a same even. Moreover, other nodes can save power because they didn’t send the same event. In our simulation, the RIET algorithm can enhance sensor nodes’ life time about 12.9 times and saving power consumption about 52.43 % than tradition algorithms.

A Transmission Power Optimization with a Minimum Node Degree for Energy-Efficient Wireless Sensor Networks with Full-Reachability

Abstract: Transmission power optimization is the most significant factor in prolonging the lifetime and maintaining the connection quality of wireless sensor networks. Un-optimized transmission power of nodes either interferes with or fails to link neighboring nodes. The optimization of transmission power depends on the expected node degree and node distribution. In this study, an optimization approach to an energy-efficient and full reachability wireless sensor network is proposed. In the proposed approach, an adjustment model of the transmission range with a minimum node degree is proposed that focuses on topology control and optimization of the transmission range according to node degree and node density. The model adjusts the tradeoff between energy efficiency and full reachability to obtain an ideal transmission range. In addition, connectivity and reachability are used as performance indices to evaluate the connection quality of a network. The two indices are compared to demonstrate the practicability of framework through simulation results. Furthermore, the relationship between the indices under the conditions of various node degrees is analyzed to generalize the characteristics of node densities. The research results on the reliability and feasibility of the proposed approach will benefit the future real deployments.

Distributed Multiple Description Coding : Principles, Algorithms and Systems

Abstract: This book examines distributed video coding (DVC) and multiple description coding (MDC), two novel techniques designed to address the problems of conventional image and video compression coding Covering all fundamental concepts and core technologies, the chapters can also be read as independent and self-sufficient, describing each methodology in sufficient detail to enable readers to repeat the corresponding experiments easily Topics and features: provides a broad overview of DVC and MDC, from the basic principles to the latest research; covers sub-sampling based MDC, quantization based MDC, transform based MDC, and FEC based MDC; discusses Sleplian-Wolf coding based on Turbo and LDPC respectively, and comparing relative performance; includes original algorithms of MDC and DVC; presents the basic frameworks and experimental results, to help readers improve the efficiency of MDC and DVC; introduces the classical DVC system for mobile communications, providing the developmental environment in detail

Similarity-based evolution control for fitness estimation in particle swarm optimization

Abstract: Evolution control in the surrogate-assisted evolutionary and other meta-heuristic optimization algorithms is essential for their success in efficiently achieving the global optimum. In order to further reduce the number of fitness evaluations, a similarity-based evolution control method is introduced into the fitness estimation strategy for particle swarm optimization (FESPSO) [1]. In the proposed method, the fitness of a particle is either estimated or evaluated, depending on its similarity to the particle whose fitness is known. The performance of the proposed algorithm is examined on eight benchmark problems, and the simulation results show that the proposed algorithm is highly competitive on reducing the number of required fitness evaluations using the computationally expensive fitness function.

Two classifiers based on nearest feature plane for recognition

Abstract: In this paper, two improved methods based on nearest feature plane (NFP), called as center-based nearest feature plane (CNFP) and line-based nearest feature plane (LNFP), are proposed for recognition. Borrowing the concept from the nearest neighbor plane (NNP) classifier and center-based nearest neighbor (CNN) classifier, the proposed methods choose the valuable representation of the class to reduce the computational complexity of NFP. At the same time, CNFP and LNFP try their best to get the better performance than NFP classifier. A large number of experiments on Yale face database and soil object database are used to evaluate the proposed algorithms. The experimental result demonstrate that the proposed method take lower computational complexity and achieve better recognition rate than the other improved classifiers.

Data Hiding Based on Subsampling and Compressive Sensing

Abstract: This paper proposes a novel scheme that considers the data hiding with sub sampling and compressive sensing. We utilize the characteristics of compressive sensing, sparsity and random projection, to embed secret data in the observation domain of the sparse image obtained through compressive sensing. The high bit correction rate (BCR) in experiments shows the high accuracy of our proposed method.

An Echo-Aided Bat Algorithm to Support Measurable Movement for Optimization Efficiency

Abstract: An Echo-Aided Bat Algorithm (EABA) based on measurable movement is proposed to improve optimization efficiency in this study. The conception is to employ the echo time to measure the distance from bats and objective. The bats emit an ultrasound to objective to measure the time of a round trip between their position and objective position. The echo time can guide the bats to correct velocity, direction and movement step. And the bats can more accurately measure the position of objective to adjust its step to find the better solution. There are many scenarios with different population sizes and objective functions to verify the performance of the proposed EABA. The experimental numeric result shows that EABA has better ability of search to improve the quality of the best solution than BA. The solution performance is improved by 45% and 30% for the functions of low complexity and high complexity in comparison with the original bat algorithm, respectively.

Hierarchical gradient diffusion algorithm for wireless sensor networks

Abstract: In this paper, a hierarchical gradient diffusion algorithm is proposed to solve the transmission problem and the sensor node's loading problem by adding several relay nodes and arranging the sensor node's routing path. The proposed hierarchical gradient diffusion aims to balance sensor node's transmission loading, enhance sensor node's lifetime, and reduce the data package transmission loss rate. According to the experimental results, the proposed algorithm not only reduces power consumption about 12% but also decreases data loss rate by 85.5% and increases active nodes by about 51.7%.

Design of a monocular simultaneous localisation and mapping system with ORB feature

Abstract: Vision-based Simultaneous Localisation and Mapping (Visual SLAM) is a new hot topic in intelligent robotic applications. A new method for the implementation of a visual SLAM system with monocular vision is proposed in this paper. The general framework of our system is first displayed, and then all the main sub-processes are described step by step. In our design we use the ORB feature to represent each natural landmark with an improved map management and modified covariance extended Kalman filter (MVEKF) to estimate the 6D pose of a free-moving camera. In order to validate and demonstrate the performance of the system, some related experiments are carried out. The experimental results show that our method is feasible, robust and efficient.

Feature Extraction Based on Nearest Feature Line and Compressive Sensing

Abstract: -In this paper, a novel feature extraction algorithmbased on nearest feature line and compressive sensing is proposed.The prototype samples are transformed to compressivesensing domain and then are performed Neighborhood discriminantnearest feature line analysis (NDNFLA) in the proposedalgorithm. This method can reduce the computational complexityfor feature extraction using nearest feature line. At the same time.its average recognition rate is very close to that of NDNFLA. Theproposed algorithm is applied to image classification on AR faceDatabase. The experimental results demonstrate the effectivenessof the proposed algorithm

Fast Image Blending and Deghosting for Panoramic Video

Abstract: The technology of panoramic image construction is widely used in many conditions. As the increasingly demand of monitoring system application, the attribute of real-time is more important during the panoramic construction. This paper describes a method of fast building panoramic video and output frames in real-time. SURF algorithm is employed to extract features and a new method is developed to eliminate matching mistakes in this work. As traditional transform method use only one homography matrix, the stitch error in detail is obvious. In this paper the modified dual-homography is proposed to improve the stitch performance. Ghosting effect is a universal problem in image stitching and difficult to avoid. This topic use image subtraction and smooth transition blending method to avoid ghost. Since this topic skips the seam finding process, so the time consumption is low and the method can achieve real-time requirement.

A new non-exact aho-corasick framework for ECG classification

Abstract: The Aho-Corasick (AC) algorithm is a popular and useful exact string matching algorithm for text searching and deep packet inspection. However, it has seldom been used for non-exact classification or identification. We propose a novel framework to make use of AC for non-exact matching in the ECG identification. The AC classification (ACC) algorithm converts ECG waveforms into several short patterns for AC, and decides the identification result by AC matched counting value. In our experiments, the results are surprisingly good and superior to previous algorithms. So, we designed an AC algorithm application for non-exact classification with high accuracy. Meanwhile, ACC inherits the advantage from AC of being capable of handling a large pattern set with linear time complexity.

Digit-Serial GNB Multiplier Based on TMVP Approach over GF(2m)

Abstract: In the four arithmetic operations of ECC, multiplicationis the most important arithmetic operation. Efficienthardware implementation of arithmetic operations, speciallymultiplication operation, over the finite field using Gaussiannormal basis is attractive. In this paper, we proposed a digitserialGaussian normal basis multiplier in GF(2m). The proposedmultiplier is based on the subquradtic Toeplitz matrixvectorproduct approach. We use Tensor product to design4-way splitting method of TMVP. The main advantage of theproposed multiplier is can be designed and implementation ina small size.

The evolutionary random interval fingerprint for a more secure wireless communication

Abstract: In this paper, we propose a novel evolutionary Random Interval Fingerprint RIF for active RFID and ZigBee systems. This new approach can enable more secure multi-party communication since, if the wireless packets are forged by another wireless communication party, the interval fingerprint can provide another way to detect the spoofing packet. Moreover, the random evolutionary algorithms, both genetic and memetic, are also proposed as a means to generate the random interval fingerprint. Compared to the conventional random generator, our approach is flexible in generating uniform random and long cycle numbers, and more robust for the anti-cracking. It is difficult for the forged party to produce the fake random intervals. Finally, we provide an application example, a completed work survey, pseudo-code and analysis result to prove that our concept is feasible for the Wireless communication.

A Novel Brain-Computer Interface Controller Scheme for Multimedia Control

Abstract: Due to the revolutionary developments in computer science and electronics, multiple kinds of new controllers are introduced Such as eye tracking, voice recognition and hand gesture recognition But these controller scheme still have their drawbacks In this paper, we proposed an improved brain-computer interface (BCI) controller based on SSVEP Our scheme provide good accuracy and comfortable user experience