Earticle

Harmony search algorithm (HSA) is an evolutionary algorithm which is used to solve a wide class of problems. HSA is based on the idea of musician's behavior in searching for better harmonies. It tries to find the optimal solution according to an objective function. HSA has been applied to various optimization problems such as timetabling, text summarization, flood model calibration. In this paper we used HSA to select an optimal subset of features that gives a better accuracy results in solving the face recognition problem. The proposed approach is compared with the standard Principal Component Analysis (PCA). A set of images that each has a face adopted from the literature is used to evaluate the proposed algorithm. The obtained results show that using HSA to select the subset of features gives better accuracy in face recognition.

The performance of speaker verification system degrades when the test segments are utterances of short duration, therefore, we investigate the use of model representing our target speaker with his close speaker and his own speech data. We propose to create a new Speaker Model who groups close speakers (CS) achieved with two clustering algorithms in Automatic Speaker Verification A.S.V. Intra and Inter speaker’s variability are two clustering algorithm used in voice module. We compare the traditional approach which uses one specific customer model (Maximum a Posteriori Adaptation) with the Close Speaker model (Customers Families).Close Speaker Model (CSM) applied only when speaker model is weak achieves 42% of equal error rate. The results demonstrate that the log likelihood of close speakers is greater than the likelihood of client speaker. The false alarm from client and CSM are closest and we are constrained to enhance speaker model.

Scrambling transformation plays an important role in information hiding application, so offering an effective evaluation method for scrambling algorithms is becoming increasingly necessary. The paper firstly analyzed the Arnold transformation process to get some universal rules about the periodicity of scrambling process, then used the improved Intersecting Cortical Model Neural Network (ICMNN) designed especially to extract 1D signatures of the original image and scrambled images which could effectively reflect the image structure changing processing. Finally L1 norm was adopted to evaluate the scrambling degree and the universal rules obtained above were used to verify the results. The experimental results showed that the proposed method could analyze and evaluate the scrambling degree efficiently and had a promising application future.

The Case Based Reasoning (CBR) is an approach of solving problem which is based on the reuse, by analogy, of past experiences called case. It is based on the retrieval and adaptation of the old solutions to the new problems. This paper presents a Bayesian adaptation-Guided Retrieval phase for a CBR applied to the diagnosis of hepatic pathologies. The main idea consists in a modelling the case base by a Bayesian Network (BN). Its are excellent tools for modelling the uncertainty in terms of their clear graphic representation as well as the conditional probabilities laws defined on a graph. We are interested to retrieval and adaptation phases. The retrieval phase consists of selecting the most similar case of log linear model by the considering Bayesian Network as a log-linear model on the simplification of the probability. The adaptation phase means modifying solutions of retrieved cases to fit the current problem. The dependence between these two phases is defined by two measures: a similarity measure and an adaptation measure. The objective of this dependence is to guarantee the retrieved case which is the easiest to adapt and improve the performance of CBR. An example of the diagnosis of the hepatic pathologies will illustrate the presented approach.

Removing speckles is a key step for accurate pupil location in an iris recognition system. This paper proposed an algorithm to detect speckles in pupil area based on 2D Gabor filters first. Then the speckles were replaced selectively with the average intensity of a fixed square region. Finally, a novel evaluation index was introduced, which improved the pupil location accuracy rate of our former pupil location algorithm. In the widely used CASIA v3.0 iris database, the pupil location accuracy rate was improved from 97.44% to 99.55%. And in several other commonly used iris test databases with less or without speckles, this method maintained the former location accuracy. The experimental results show that our algorithm has satisfactory performance, robustness and versatility.

This paper mainly proposes notions and methods of attribute reduction in concept lattices based on rough set theory. Using dependence space of concept lattices, we first discuss the relationships between congruence relations and the corresponding concept lattices. We then define notions of attribute reduction in a formal context based on congruence relations which is to find the minimal attribute subsets preserving the congruence partition. Finally, we define discernibility matrices and Boolean functions of a formal context to calculate all attribute reducts and analyze attribute characteristics. Using this notion of attribute reduction, methods, results as well as their proof about attribute reduction in a formal context can be derived directly by those in rough set theory. Furthermore, we prove that the attribute reducts proposed in this paper also preserve all extents of formal concepts and their original hierarchy in the concept lattice.

Estimating the job cycle time is very important for the control of a factory. However, the uncertainty in the job cycle time is not easy to deal with. In order to effectively estimate the job cycle time in a wafer fabrication factory, a group of distributed sensors are used in this study. In the proposed methodology, each sensor monitors the factory conditions, and uses a fuzzy neural network to estimate the job cycle time, based on its local observation. Each sensor communicates its view and estimates to other sensors with the aid of the central control unit. According to the experimental results, the aggregate estimation performance was considerably improved through the sensors’ collaboration.

Most of the previous works of network visualization analysis depends on node graph and adjacency matrix in connection with an aspect of network topology. However, the representing node and matrix are difficult to understand the relationship between network nodes, because the node’s interaction on network is presented by a complicated node graph. In order to overcome this limitation, this paper proposes a new visualization method to represent hierarchy relationship of network nodes for analysis of data network. The proposed method uses data node correlation to construct hierarchy node relationship which can intuitively understand node interaction. Besides, it can focus on node relation on network which is modeled using node activities of data network by visualizing relationship among the internal relation of network reflecting node and external relation of network nodes.

A logic reasoning model based method is proposed in this paper for object tracking in compressed domain. First, the object is defined by the motion detector. Then, the supposed target trajectory and predicted position are estimated by the direction angle and the intensity within motion field. The possible conditions of lost, stop or occlusion are determined by logic and reasoning analysis and handled by the direction angle and the intensity in motion field. The outperformance of our method on reducing loss rate and enhancing the trajectory fit has been demonstrated by the experiments. The accurate object tracking result of the proposed method is presented as well.

One of key goals of disaster response and relief robots is to acquire knowledge about the situation that would be either dangerous or inaccessible to human operator. However, the robot is difficult to acquire tele-operations from the operator. A disconnected communication link and no-visual control situation are frequently happened in disaster areas. Moreover, disaster scenarios are typically spatially distributed, so we may need an intelligent robot that has functions such as autonomy, cooperation, and collective behaviors. Thus, we propose the bio-inspired modular robot named as ARTHROBOT so as to support emergency responders. ARTHROBOT can assemble or disassemble process based on the proposed mobile algorithms.

A new architecture for decimating finite impulse response filters is proposed. The architecture is based on using a number of accumulators; each one accumulates a partial sum corresponding to a unique set of D filter coefficients into the filter output, where D is the decimation factor. In the new decimating filter, the accumulated result of an accumulator is passed to another accumulator once for each period of D input samples, except for that of the last accumulator whereby the filter output is obtained. The size of each accumulator can be minimized, depending on the filter coefficients. A demonstrative FPGA implementation shows that this architecture is more favorable than the widely used polyphase architecture, as it requires much less area at similar power consumption.

A novel cultural quantum-behaved particle swarm optimization algorithm (CQPSO) is proposed to improve the performance of the quantum-behaved PSO (QPSO). The cultural framework is embedded in the QPSO, and the knowledge stored in the belief space can guide the evolution of the QPSO. 15 high-dimensional and multi-modal functions are employed to investigate the proposed algorithm. Numerical simulation results demonstrate that the CQPSO can indeed outperform the QPSO.

An autonomic computing system has four basic characteristics, namely self-configuration, self-optimization, self-healing and self-protection. Autonomic computing can be viewed as a new computing paradigm and it is becoming a hot research topic in distributed and ubiquitous computing area. In this paper, we not only discuss the four basic aspects of autonomic computing comprehensively based on our own understanding but also proposed autonomic agent based power-aware routing approach for ubiquitous sensor networks which is a distributed and localized routing approach. Besides, we provide an application scenario to the sensor network among which the power consumption is one of the most critical issues. The amount of agent is carefully selected and network performance such as packet delivery rate and power consumption is also compared in the simulation part.

To illustrate the different kinetics of cellular self-repair mechanism under external perturbations from outer environment, a mathematical model of DNA damage repair process is proposed by using the Kinetic Theory of Active Particles (KTAP) framework. The profile of cellular self-repair process is represented by two sub-populations, each of which is made up of the active particles with different discrete states. The dynamic kinetics of DNA damage generation, repair mRNA transcription, Repair Protein (RP) translation, DSBC synthesis are investigated by the particle interactions between the molecular pairs within DNA and RP sub-systems.

In traditional text clustering, documents appear terms frequency without considering the semantic information of each document (i.e., vector model). The property of vector model may be incorrectly classified documents into different clusters when documents of same cluster lack the shared terms. Recently, to overcome this problem uses knowledge based approaches. However, these approaches have an influence of structure of document set and a cost problem of constructing ontology. In this paper, we propose a text clustering method using semantic terms for clustering label and term weights. The semantic terms of clustering label can well express the internal structure of document clusters using non-negative matrix factorization (NMF). It can also improve the quality of text clustering which uses the term weights by WordNet. The experimental results demonstrate that the proposed method achieves better performance than other text clustering methods.

The preliminary analysis of the direct economic losses of meteorological disasters in the nearly 20 years, using the C-D production function and the factor of meteorological disasters, to measure the overall impact of meteorological disasters on the national economy. Using the 1989-2008 meteorological disasters data with economic and social statistics information established a production function model including meteorological disasters factor. The results showed that meteorological disasters have brought a loss of 9.72% to the national economy each year , is nearly four times of the average annual direct economic losses to GDP ratio (2.44%); meteorological disasters of the best and the worst year for GDP fluctuating between 11.5% -14.5% , meteorological disasters impact on the national economy is huge.

The sheer volume of new malware found each day is growing at an exponential pace. Centralized systems that collect all malware samples to central severs can cause problems of single point of failure as well as processing bottlenecks. Previous works on distributed and scalable malware analysis are mainly applied for specific or simple malware. This paper presents CCS, a collaborative online malware analysis system which is applied for various malware and well scalable. Each sensors in CCS analysis their own malware samples accurately in-situ and then CCS aggregates those analyses among sensors in a load-balance way. We implemented a proof-of-concept version of CCS and performed experiments with 917 real-world malware samples; preliminary results from our evaluation confirm that CCS has comparable performance with centralized system, but much better scalability, and is approximately consistent with the result of AV scanners.

Image registration is a necessary pre-processing step before quantitative analysis of brain MR data. A novel variational optical flow approach for image registration is proposed in this paper. The advantages of our method are as follows. We coupled bias correction and optical flow image registration within the unified variational framework. We could recover the corrected target image through the estimated bias field. Experiments on synthetic and real brain images demonstrate the advantages of our method.

The tracking and forecasting of satellite cloud images are very important in term of satellite cloud images used in weather forecast. An approach on automatic tracking of multi-target cloud cluster based on VFC Snake model is proposed on the basis of contour extraction and analysis of cloud cluster, this method can automatically acquire the new location of the target cloud cluster at each moment. A specialized detection algorithm is designed to correct the Snake's tracking results, and more accurate contour curves are obtained. For the forecasting of cloud images, the target cloud cluster’s displacement obtained in the tracking process is integrated into the cross-correlation matching to improve the matching accuracy of the cross-correlation method, and more accurate cloud motion vectors are obtained. The experimental results show that the tracking based on contour detection and analysis is fast and highly accurate, and the evolution process of cloud cluster can be directly obtained in a period of time (eg, split, merge, die and newborn), preferable results has also been made in forecasting.

When applied to wire ice-covering forecasting, the back propagation (BP) neural network is a lack of guidance for selecting the neural network initial connection weight and network structure, which contributes to the problem of a high degree of randomness and poses a difficulty for selecting an initial node with global properties. Combination traditional forecasting methods of Mean Generating Function-optimal subset regression (MGF-OSR), this paper proposes a new hybrid MGF-OSR-BP model based on Genetic Algorithm (GA) evolution BP. This paper uses the hybrid MGF-OSR-BP model based on GA evolution BP to analyze 108-ten days of ice thickness data from Erlang Mountain glacial stage, China, from 2001 to 2009.The results show that the model has a better forecast accuracy and high convergence .This paper can serve as a reference for similar middle-and long-term forecast research based on elements of time series data.

To understand the other person’s emotion, we should know the situations in which the person is surrounded and the personality of the person. In most previous studies, however, these important characteristics don’t be considered, and emotion recognition has been considered as a problem of classifying texts. In this paper, we attempt to novel approaches to utilize situational information and personality of emotional subject. We propose the method extracting situational information, and the personalized emotion model for reflecting personality of emotional subject. To extract and utilize situational information, we propose situation model using lexical and syntactic information. In addition, To reflect personality of emotional subject, we propose personalized emotion model using KBANN(Knowledge-based Artificial Neural Network). Experimental results show that the proposed system can recognize emotions more accurately and intelligently than previous text-based emotion recognition systems.

Attacking on inter-domain routing system degrades the availability and performance of Internet severely. It is challenge to defend against the extreme attacks which exhaust the resources of routers by generating a great number of update messages. In this paper, we propose two mechanisms to protect Internet from such attacks: to isolate attacks in local region, unnecessary updates are suppressed without affecting the correctness of routing; to break down the route flapping which repeatedly generates updates, the paths selected are validated to detour the attacked links, which diffuses the deliberately attacks to random attacks . Simulation shows our methods greatly decrease the number of updates under such attacks, and isolate the attacks in local region of network.

Recently, Cloud Computing introduces some new concepts that entirely change the way applications are built and deployed. Usually, Cloud systems rely on virtualization techniques to allocate computing resources on demand. Thus, scalability is a critical issue to the success of enterprises involved in doing business on the cloud. In this paper, we will describe the novel virtual cluster architecture for dynamic scaling of cloud applications in a virtualized Cloud Computing environment. An auto-scaling algorithm for automated provisioning and balancing of virtual machine resources based on active application sessions will be introduced. Also, the energy cost is considered in the proposed algorithm. Our work has demonstrated the proposed algorithm is capable of handling sudden load requirements, maintaining higher resource utilization and reducing energy cost.

Research on periodic pattern mining has gained a great attention in the past decade. Periodic pattern mining discovers valid periodic patterns in a time-related dataset. This study proposed an efficient 2-D linked structure and the OEOP (One Event One Pattern) algorithm to discover all kinds of valid segments in each single event sequence. Then, this study combines these valid segments found by OEOP into 1-patterns with multiple events, and multiple patterns with multiple events periodic patterns. The experimental results show that the proposed algorithm has good performance and scalability.

The modeling and verification of real-time systems is a challenging task in the area of software engineering. This paper proposes a formal method for modeling and verification of real-time systems based on aspect-oriented timed statecharts and linear-time temporal logic. Behaviors of real-time systems are modeled by aspect-oriented timed statecharts, while key properties of systems are specified by linear-time temporal logic. Moreover, aspect-oriented timed statecharts are translated to timed automata with guards to simulate the executable paths of systems and model checking technologies are applied to the verification of models. An elevator example illustrates our modeling and verification method.

One of key challenges in deploying large-scale Wireless Sensor Network (WSN) is system longevity. In this paper we propose an harvesting aware system of SMSN (Sustainable Mobile Sensor Network) as a way to extend system lifetime. The mobile sensor nodes visit remote energy station to recharge battery when node’s energy drains below a threshold level. Our framework consists of energy model, motion control system and data transfer protocol. The simulation result shows the sustainability of our SMSN. Our results along with simulation can be used for certain guideline for realistic development of such systems.

This paper proposes a hybrid query expansion method named GAO, which derives from the fact that more and more documents have been annotated with one or several ontology concepts based on their semantic. The GAO method employs a combination of global analysis and ontology technology to improve query expansion performance. The global analysis technology is used to obtain term-concept association, and ontology technology is used to carry out semantic reasoning. Experimental results of query expansion on two different corpuses show that, compared with traditional query expansion methods, the GAO method can improve the precision effectively.

Trust is part of our daily life and thus can be used as a mechanism for providing security in computer networks. In this paper, by considering the dynamic nature of trust, especially the temporal and spatial characteristics in security of society, we describe trust from four factors: access time, place, history behavior and risk control strategy and then apply fuzzy clustering method and information entropy theory to design a weight allocation algorithm for these factors to compute a value for trust.

In distributed storage system, storage service provider (SSP) is a critical component to receive and response these requests from clients. We propose a real-time and flexible management of SSP in this paper. With the advantages from virtualization and cloud computing, the states of SSP main VM are observed by another isolated virtual machine, and the administrator can inspect the storage service provider through the uniform management interface in real-time.

A face recognition (FR) system is automatically identifying or verifying a personal face acquired from a digital camera or a image generation device. In order to do this, facial features from the acquired image should be extracted and compared with a facial database. All FRs face an obstacle related to the viewing angle of the face including poor lighting and low resolution. Because of those problems, its recognition rate substantially decreases. In this paper, a newly weighted regularization parameter based FR system which can improve recognition rate under certain environmental constraints is proposed. This approach is based on the conventional regularized linear discriminant analysis (R-LDA) and includes Artificial Neural Network (ANN) which can improve face recognition rate with a prominent classification ability. The revised R-LDA algorithm is attempted to address the Small Sample Size (SSS) problem that encountered in all FRs and the ANN is useful to detect the frontal views of faces. This algorithm has been tested over 350 images (35 classes) of Olivetti Research Lab (ORL) database using MATLAB. Its test results give us recognition rates of above 95%. In addition, it is also tested on the mirror and combination of the ORL database and the recognition performances are shown that the system is fairly robust and has the performance of more than 90%.