Advances in Swarm Intelligence

In recent years, there has been a growing interest in algorithms inspired by the behaviors of natural phenomena. However, the performance of any single pure algorithm is limited by the size and complexity of the problem. To further improve the search effectiveness and solution robustness, hybridization of different algorithms is a promising research direction. In this paper, we propose a hybrid iteration algorithm by combing the gravitational search algorithm with the clonal selection. The gravitational search performs exploration in the search space, while the clonal selection is implemented to carry out exploitation within the neighborhood of the solutio found by gravitational search. The emerged hybrid algorithm, called GSCSA, thus reasonably combines the characteristics of both base algorithms. Experimental results based on several benchmark functions demonstrate the superiority of the proposed algorithm in terms of solution quality and convergence speed.

By changing the linear encoding and redefining the evolving rules, particle swarm algorithm is introduced into genetic programming and an hybrid genetic programming with particle swarm optimization (HGPPSO) is proposed. The performance of the proposed algorithm is tested on tow symbolic regression problem in genetic programming and the simulation results show that HGPPSO is better than genetic programming in both convergence times and average convergence generations and is a promising hybrid genetic programming algorithm.

The traditional Cellular Automation-based

Physarum

model reveals the process of amoebic self-organized movement and self-adaptive network formation based on bubble transportation. However, a bubble in the traditional

Physarum

model often transports within active zones and has little change to explore new areas. And the efficiency of evolution is very low because there is only one bubble in the system. This paper proposes an improved model, named as Improved Bubble Transportation Model (IBTM). Our model adds a time label for each grid of environment in order to drive bubbles to explore new areas, and deploys multiple bubbles in order to improve the evolving efficiency of

Physarum

network. We first evaluate the morphological characteristics of IBTM with the real

Physarum

, and then compare the evolving time between the traditional model and IBTM. The results show that IBTM can obtain higher efficiency and stability in the process of forming an adaptive network.

This paper addresses the solution of weighted set covering problems using cultural algorithms. The weighted set covering problem is a reasonably well known NP-complete optimization problem with many real world applications. We use a cultural evolutionary architecture to maintain knowledge of diversity and fitness learned over each generation during the search process. The proposed approach is validated using benchmark instances, and its results are compared with respect to other approaches which have been previously adopted to solve the problem. Our results indicate that the approach is able to produce very competitive results in compare with other algorithms solving the portfolio of test problems taken from the ORLIB.

The Concerning the problem of near space target intercepting, double-goal optimization model is established in order to maximize the response pace and to minimize the engine consumption at the same time. Impulse engine ignition algorithm based on Genetic Particle Swarm Optimization is presented. The global optimal solution of the optimal problem could be obtained fast by this algorithm. Simulation results indicate that the Ignition control algorithm can obtain high control precision with low engine consumption.

This paper shows that learning by imitation leads to a positive effect not only in human behavior but also in the behavior of the autonomous agents (AA) in the field of self-organized creation deposits. Indeed, for each agent, the individual discoveries (i.e. goals) have an effect on the performance of the population level and therefore they induce a new learning capability at the individual level. Particularly, we show through a set of experiments that adding a simple imitation capability to our bio-inspired architecture allows increasing the ability of agents to share more information and improving the overall performance of the whole system. We will conclude with robotics’ experiments which will feature how our approach applies accurately to real life environments.

In this paper, an indexed K-D tree is proposed to solve the problem of neighborhoods generation in swarm robotic simulation. The problem of neighborhoods generation for both robots and obstacles can be converted as a set of range searches to locate the robots within the sensing areas. The indexed K-D tree provides an indexed structure for a quick search for the robots’ neighbors in the tree generated by robots’ positions, which is the most time consuming operation in the process of neighborhood generation. The structure takes full advantage of the fact that the matrix generated by robots’ neighborhoods is symmetric and avoids duplicated search operations to a large extent. Simulation results demonstrate that the indexed K-D tree is significantly quicker than normal K-D tree and other methods for neighborhood generation when the population is larger than 10.

Consumer’s behavior can be modeled using a utility function that allows for measuring the

success

of an individual’s decision, which consists of a tuple of goods an individual would like to buy and the hours of work necessary to pay for this purchase and consumption. The success of such a decision is measured by a utility function which incorporates not only the purchase and consumption of goods, but also leisure, which additionally increases the utility of an individual. In this paper, we present a new agent based social simulation in which the decision finding process of consumers is performed by Particle Swarm Optimization (PSO), a well-known swarm intelligence method.

PSO appears to be suitable for the underlying problem as it is based on previous

and

current information, but also contains a stochastic part which allows for modeling the uncertainty usually involved in the human decision making process. We investigate the adequacy of different bounding strategies that map particles violating the underlying budget constraints to a feasible region. Experiments indicate that one of these bounding strategies is able to achieve very fast and stable convergence for the given optimization problem. However, an even more interesting question refers to adequacy of these bounding strategies for the underlying social simulation task.

In the field about multi-agent system, the payoff rationality is an important factor for the forming of a multi-agent coalition structure. In this paper, we regard a payoff vector belonging to the bargaining set in classical cooperative game as a stable payoff vector of multi-agents. Then, we propose an approach to find the stable payoff vector based on genetic algorithm. Finally, the experimental results and analysis are showed about success rates and running times of our proposed approach.

The clustering method based on one-class support vector machine has been presented recently. Although this approach can improve the clustering accuracies, it often gains the unstable clustering results because some random datasets are employed for its initialization. In this paper, a novel initialization method based on the core clusters is used for the clustering algorithm based one-class support vector machine. The core clusters are gained by constructing the neighborhood graph and they are regarded as the initial datasets of the clustering algorithm based one-class support vector machine. To investigate the effectiveness of the proposed approach, several experiments are done on four datasets. Experimental results show that the new presented method can improve the clustering performance compared to the previous clustering algorithm based on one-class support vector machine and k-means approach.

Remote sensing is an important tool in a variety of scientific researches which can help to study and solve many practical environmental problems. Classification of remote sensing image, however, is usually complex in many respects that a lot of different ground objects show mixture distributions in space and change with temporal variations. Therefore, automatic classification of land covers is of practical significance to the exploration of desired information. Recently, support vector machine (SVM) has shown its capability in solving multi-class classification for different ground objects. In this paper, the extension of SVM to its online version is employed for terrain image classification. An illustration of online SVM learning and classification on San Francisco Bay area is also presented to demonstrate its applicability.

Model optimization namely the kernel function and parameter selection is an important factor to affect the generalization ability of support vector machine (SVM). To solve model optimization problem of support vector machine classifier, a novel algorithm (GC-ABC) is proposed which integrate artificial bee colony algorithm, greedy algorithm and chaos search strategy. The simulation results show that the accuracy of SVM optimized by GC-ABC is superior to the SVM optimized by genetic algorithm and ant colony algorithm. The experiments further suggest that GC-ABC algorithm has fast convergence and strong global search ability, which improves the performance of the support vector machine.

An algorithm to solve the least square support vector machine (LSSVM) is presented. The underlying optimization problem for LSSVM follows a system of linear equations. The proposed algorithm incorporates a fuzzy c-mean (FCM) clustering approach and the application of a recurrent neural network (RNN) to solve the system of linear equations. First, a reduced training set is obtained by the FCM clustering approach and used to train LSSVM. Then a gradient system with discontinuous righthand side, interpreted as an RNN, is designed by using the corresponding system of linear equations. The fusion of FCM clustering approach and RNN overcomes the loss of spareness of LSSVM. The efficiency of the algorithm is empirically shown on a benchmark data set generated from the University of California at Irvine (UCI) machine learning database.

The locality sensitive k-means clustering has been proposed recently. However, it performance depends greatly on the choice of the initial centers and only proper initial centers enable this clustering approach to produce a better accuracies. In this paper, an evolutionary locality sensitive k-means clustering method is presented. This new approach uses the genetic algorithms for finding its initial centers by minimizing the Davies Bouldin clustering validity index regarded as the fitness function. To investigate the effective of our approach, some experiments are done on several datasets. Experimental results show that the proposed method can get the clustering performance significantly compared to other clustering algorithms.

Knowledge discovery in social networks is not a trivial task. Often research in this context uses concepts of data mining, social network analysis, trust discovery and sentiment analysis. The connected network of people is generally represented by a directed graph (social graph), whose formulation includes representing people as nodes and their relationships as edges, which also can be labeled to describe the relationship (eg. friend, son and girlfriend). This environment of connected nodes behaves like a dynamic network, whose nodes and connections are constantly being updated. People tend to communicate or relate better with other people who have a common or similar way of thinking, which generates groups of people with common interests, the communities. This paper studies the use of a graph clustering approach, the Coring Method, originally employed in image segmentation task, in order to be applied in the context of community discovery on a social network environment.

Rival penalized competitive learning (RPCL) has provided attractive ways to perform clustering without knowing the exact cluster number. In this paper, a new variant of the rival penalized competitive learning is proposed and it performs automatic clustering analysis of seismic data. In the proposed algorithm, a new cost function and some parameter learning methods will be introduced to effectively operate the process of clustering analysis. Simulations results are presented showing that the performance of the new RPCL algorithm is better than other traditional competitive algorithms. Finally, by clustering the seismic data, a kind of geological characteristic, underground rivers, can be extracted directly from the 3D seismic data volume.

It is important to analyze and predict the measuring data trend in industrial measuring and controlling process. The paper introduces a method for predicting the trend of the current measuring data based on clustering the historical data. It calculates the similarities of the current trend and the bases result from the clustering. And with these similarities, the future trend of the current measuring data can be predicted , the combination of the above bases representing low frequency and a reviser representing high frequency. The simulation shows the weights of high or low frequency have effect on the precision of predict results. It is also found that the proposed method can predict more precisely than the RBFNNs method in high frequency.

Measuring micro-blogging user influence is very important both in economic and social fields. In this paper, we propose a user-tweet interaction model to describe the relationships among users and tweets. Considering the time affect,

TAC

(time-effectiveness attenuation coefficient) is proposed when calculating tweet influence which consists of retweet influence and comment influence. Then we make a detail analysis on the generation of user influence which consists of post influence and follow influence based on the results of tweet influences. We also discuss the correlation between post influence and follow influence by use of Spearman’s rank correlation coefficient. At last, we rank users by calculating the bias spatial distances. Taking Sina micro-blogging as background, after a series of experiments, we believe that our method is accurate and comprehensive when measuring the influences of micro-blogging users.

Community leaders are individuals who have huge influence on social network communities. Discovering community leaders in social networks is of great significance for research on the structures of the social networks and for commercial application. Based on the core idea of the PageRank algorithm, this paper firstly processes data selected from

Sina

microblog, and extracts three key indicators, comprising the number of followers, the number of comments and the number of reposts; then based on their mutual relationship, that is following or followed, it obtains the weight of influence for each individual user; and then after a finite number of iterations, this paper identifies the community leader in

Sina

microblog, by which its comprehensive influence on its community are reflected.

In this paper, we introduce an effective machine learning method which can capture the temporal causal structures between irregular time series to forecast China GDP growth rate with Mixed Frequency data set. The introduced method first generalized the inner product operator via kernels so that regression-based temporal casual models can be applicable to irregular time series, then the temporal casual relationships among the irregular time series are studied by Generalized Lasso Granger (GLG) graphical models. The main advantage of this approach is that it does not directly estimate the values of missing data of low frequency time series or has restricted assumptions about the generation process of the time series. By applying this method to a 17 macroeconomic indicators GLG model, the forecasting accuracy is better than the autoregressive (AR) benchmark model and a widely used mixed-data sampling (MIDAS) model.

The purpose of this paper was to provide help for poison identification via the Bayesian network according to the observed preliminary symptoms of the poisoning people. We proposed a novel improvement on K2 algorithm to solve the problem of the lack of data under the special context. Determining initial node sequence of K2 algorithm via Markov blanket, we improved greatly Bayesian network structure learning with small datasets. Bootstrap data expansion and Gibbs data correction combining with maximum weight spanning tree (MWST) were used to expand the original small data set to further improve the performance and reliability of the structure learning. Then we applied this kind of combination scheme into a real data set to verify its validity and reliability. Finally we were able to quickly distinguish between a variety of biochemical reagents with this method, and the result of the inference can be used to guide emergency rescue after certain biochemical terrorism attack.

The idea of cluster ensemble is combining the multiple clustering of a data set into a consensus clustering for improving the quality and robustness of results. In this paper, a new text clustering ensemble (TCE) algorithm is proposed. First, text clustering results of applying k-means and semantic sequence algorithms are produced. Then in order to generate co-association matrix between semantic sequences, the clustering results are combined based on the overlap coefficient similarity concept. Finally, the ultimate clusters are obtained by merging documents corresponding to similar semantic sequence on this matrix. Experiment results of proposed method on real data sets are compared with other clustering results produced by individual clustering algorithms. It is showed that TCE is efficient especially on long documents set.

This paper applies B-Cell algorithm (BCA) for credit scoring analysis problems. The proposed BCA-based method is combined with k-nearest neighbor (kNN) classifiers. In the algorithm, BCA is introduced to select the optimal feature subsets and kNNs are used to classify the investors in different groups representing different levels of credit in the classification phase. Experiments employing the benchmark data sets from UCI databases will be used to measure the performance of the algorithm. Its comparison with genetic algorithm, particle swarm optimization and ant colony optimization will be shown.

A new text categorization algorithm based on Hidden Markov Model is proposed. At first, semantic clusters are obtained from training data set. The association between semantic clusters is modeled as Hidden Markov Model. Combining with the forward algorithm, the strategy could realize automatic text categorization. From the simulation, the proposed text categorization algorithm is better in categorization precision. Moreover, it works well independent of the number of considered categories compared to the priori art algorithms.

In this paper, we propose a reversible data hiding technique to improve Zhang and Hong et al.’s methods on cartoon images. Zhang and Hong et al. exploit the block complexity for data extraction and image recovery. Their methods are efficient for natural images, however, the results are unsatisfactory when applies on cartoon images consisting of large flat area. By unbalanced flipping the bits of pixel groups, the block complexity before and after flipping can be distinguished and thus the error rate can be further reduced. Experimental results show that the proposed method has lower error rate than those of Zhang and Hong et al.’s methods without degrading the image quality.

As the computer aided design becomes widely used, the copyright violation of two-dimensional CAD engineering graphics urgently needs to be studied. But the current watermarking algorithms have many flaws such as lack of robustness, watermark capacity limitation, and so on. A new algorithm which based on DCT transformation and chaos system is proposed in this paper. The algorithm adopts DCT in watermark preprocessing to reduce the watermark information redundancy, and then classifies the entities of their handles through the chaos system to ensure the security of watermark. The watermark is embedded by modifying entity’s line width slightly utilizing HVS characteristics. Experimental results show that the proposed algorithm has good imperceptibility; it’s robust against operations such as translation, rotation, scaling, entity addition/deletion and combination of these attacks. The algorithm embeds the meaningful image into 2D CAD engineering graphic; it’s useful in practical applications.

The ship’s security depends on the patrol, which is difficult to ensure ship safety in real-time. In order to secure ship more safety, the intelligent video surveillance technology is applied to the ship. Firstly, the background model is established through codebook algorithm, then the movement target of the human are detected accurately. Secondly, the characteristics of the human body are extracted through HU invariant moments. By similarity matching with the abnormal behavior template and the feature of aspect ratio of the human body, the abnormal behavior of the human body is detected. Finally, the experimental results show that this algorithm can be achieved very well. It has obvious advantages on frame difference algorithm and mixed Gaussian algorithm. In the actual environment of anchoring ship, the abnormal behavior of the human body is detected effectively.

For the purpose of discovering security flaws in software, many dynamic and static taint analyzing techniques have been proposed. The dynamic techniques can precisely find the security flaws of the software; but it suffers from substantial runtime overhead. On the other hand, the static techniques require no runtime overhead; but it is often not accurate enough. In this paper, we propose HYBit, a novel hybrid framework which integrates dynamic and static taint analysis to diagnose the security flaws for binary programs. In the framework, the source binary is first analyzed by the dynamic taint analyzer; then, with the runtime information provided by its dynamic counterpart, the static taint analyzer can process the unexecuted part of the target program easily. Furthermore, a taint behavior filtration mechanism is proposed to optimize the performance of the framework. We evaluate our framework from three perspectives: efficiency, coverage, and effectiveness, and the results are encouraging.

Each year bridge collapse causes huge loss in China. The damage identification of bridges is a difficult problem. The Pattern recognition is an important method in security assessment of structural health monitoring. Taking a railway bridge as an example, the paper introduces the application of the pattern recognition algorithms in damage identification. It is concluded that preparation work involved infinite element analysis, feature extract, and sample training is important to improve the identification effect for the pattern recognition.

Given the increasing number of users of computer systems and networks, it is difficult to know the profile of the latter and therefore the intrusion has become a highly prized of community of network security. In this paper to address the issues mentioned above, we used the data mining techniques namely association rules, decision trees and Bayesian networks. The results obtained on the KDD’99 benchmark has been validated by several evaluation measures, and are promising and provide access to other techniques and hybridization to improve the security and confidentiality in the field.

This paper presents a model called Brownian snake measure-valued Markov decision process (BSMMDP) that can simulate an important characteristic of human thought, that is, when people think problems, sometimes they can suddenly connect events that are remote in space-time so as to solve problems. We also discuss how to find an (approximate) optimal policy within this framework. If Artificial Intelligence can simulate human thought, then maybe it is beneficial for its progress. BSMMDP is just following this idea, and trying to describe the talent of human mind.

For the necessary of low-power data transmission in wireless sensor networks (WSN), this work is main for the data characteristics’ analysis of WSN and so we have create a data model based on note load method. And finally we get the energy consumption of the sensor network model and its data transmission delay model. We have report the network lifetime maximization solution algorithm on the premise of ensure the application delay requirements. The theoretical analysis and simulation results shown that the method can effectively extend the network lifetime.

Through analyzing the type of infected population, a contact model with covert infection is introduced. The stability of this model in small world network is studied according to the property of contact network. To the difference of immune pattern, the effectiveness and feasibility of the proposed model is validated.

In this paper, we present to utilize Genetic Algorithms (GAs) as tools to model control processes. Two different crossover operators are combined during evolution to maintain population diversity and to sustain local improvement in the search space. In this manner, a balance between global exploration and local exploitation is reserved during genetic search. To verify the efficiency of the proposed method, the desired control sequences of a given system are solved by the optimal control theory as well as GA with hybrid crossovers to compare their performances. The experimental results showed that the control sequences obtained from the proposed GA with hybrid crossovers are quite consistent with the results of the optimal control.

With the development of various advanced sensors, and some sensing technologies are not mature, so that measurement information was being uncertain, incomplete. This paper adopts an intelligent fusion algorithm with Rough Set for reduction of the attribute set and target set for the raw data from various sensors. Consequently the noise and redundancy will be reduced in sampling. Then constructs information prediction system of SVM according to the preprocessing information structure, and solves the problem of multisensor data fusion in the situation of small sample and uncertainty. In order to get the optimal fusion accuracy, it uses PSO for fusion parameters. To make operation faster and increase the accuracy of the fusion, a feature selection process with PSO is used in this paper to optimize the fusion accuracy by its superiority of optimal search ability.

In order to overcome the effect of blocking in process of target tracking under stationary camera, a target tracking algorithm based on online Adaboost was presented. Codebook model was set up to detect moving target in YUV color space; in process of tracking, feature of online Adaboost fused texture contours and color, then accurate target location was obtained. The experimental results show that, the detecting algorithm in this paper has good detecting results, which provides assistance to tracking. The proposed tracking algorithm is effective for the targets having blocking, even a large area of blocking in more complex scenes.

Wireless sensor networks are usually deployed in complex environments; an attacker can easily inject false data by capturing nodes, causing serious consequences. The main work of this paper is as follows. Firstly, the logical loop model is created based on the estimated value of the source events of every wireless sensor network node. Secondly, each node based on RSSI find neighbor nodes by establishing consistency loop model. Finally, the malicious node is determined by comparing the similarities and differences of the nodes between the two loop models. The simulation shows that this mechanism is effective to inhibit the infringement of malicious nodes to the network, and improve network security performance.

Cell motion analysis contributes to research the mechanism of the inflammatory process and to the development of anti-inflammatory drugs. This paper aims to develop an accurate and robust algorithm to track multiple colliding cells and further characterize the dynamics of each cell. First, a hybrid cell detection algorithm is proposed to obtain reliable measurements in cell collision images. Second, a variant of interacting multiple models particle filter is designed for analysis of cell motion behaviors. The simulation results show that our algorithm could obtain favorable performance compared with other methods.

Existing localization algorithms didn’t consider the important factor of the antenna measuring angles. And most wireless indoor localization algorithms require a site survey process which is time-consuming and labor-intensive. This paper presents a featured region localization algorithm without site survey and discusses the measured angle in different intervals. According to the relationship among the fingerprints sets of the same angle interval, our proposed algorithm is used to find the featured points within the region. Regions of points are determined by calculating Euclidean distance between the points and APs (Access Points). Experiments are conducted in an 8m by 8m laboratory, and results show that this algorithm has superior performance compared with existing algorithms.

Two critical issues of designing scalable routing algorithm in ad hoc networks are that it should be robust to frequent path disruptions caused by node mobility and limit routing overhead. This paper argues grouping nodes with the motion direction of source to limit the range of RREQ broadcast. This kind of grouping also ensures that nodes, belonging to the same group, are more likely to establish longtime existent paths as they have similar motions. During route discovery, only the nodes which are in same group with source are allowed to rebroadcast RREQ. So the destination can select a most stable route to reply. The performance is evaluated through computer simulation with NS2. Simulation results indicate the proposed routing algorithm can limit routing overhead and enhance route stability effectively.

RoboCup Soccer offers a challenging platform for intelligent soccer agents to continuously perceive their environment and make smart decisions autonomously. During a soccer match, once a robot takes possession of the ball, the most important decision it has to make is to plan a route from its current location to opponents’ goal. This paper presents an artificial neural network based approach for path planning. The proposed approach takes the current state of the environment as an input and provides the best path to be followed as an output. The weights of the neural network have been optimized using three computational intelligence based techniques, namely evolutionary algorithms (EA), particle swarm optimization (PSO), and artificial immune system (AIS). To assess the performance of these approaches, a baseline search mechanism has been suggested that works on discrete points in the solution space of all possible paths. The performance of the base line and the neural networks based approach(es) is compared on a synthetic dataset. The results suggest that the neural network evolved via PSO based approach performs better than the other variations of neural networks as well as the baseline approach.

P systems are biologically inspired theoretical models of distributed and parallel computing. Hamilton path is a classical NP problem, recently, there are lots of methods to solve it. Today we give a new and efficient algorithm to this classic. This paper uses the improved P system with priority and promoters/inhibitors to give an efficient solution to Hamilton path problem. We give two examples to illustrate our method’s feasibility. We discuss future research problems also.

With the development of information technology, many new manufacturing models, e.g. virtual manufacturing, cloud manufacturing, have emerged to enhance interoperability and collaboration among the manufacturing enterprises. As a dominate factor for production performance, manufacturing capability has attract great attention and needs to be well analyzed and measured to assist the decision making in manufacturing process. A number of works have been carried out for measuring manufacturing capability. However, most of them modeled manufacturing capability in a static manner, without considering the dynamic characteristics as well as the online and real-time collected manufacturing operation data. In this paper, we analyze the components of manufacturing capability and propose a dynamic comprehensive evaluation method of manufacturing capability for a job shop, using subjective and objective analysis. A case study is presented and the results demonstrate that it is effective and flexible to evaluate the manufacturing capability for a job shop.

Aimed at the problem of the aviation swarm convoy and transportation mission, in this paper, the swarm transport path planning method and the swarm behavior control method were put forward. Taking all kinds of static constraints into account, these methods realized the path planning in the planning space, achieved the swarms’ evasion to the fixed threat. The track data smooth processing was done so that it can be used in the fly. The methods of swarms’ behavior control can realize the evasion to the emergent threat or obstacles and response real-timely to the coming attacker target. The simulation results show that the swarm transport path planning method and the swarm convoy behavior control method are feasible. It can meet the needs of aviation swarm convoy and transportation mission.

Residential gateways that interconnect home networks, pubic networks, and smart household devices play a critical role in intelligent home systems. However, the existing gateways could hardly be adapted to emerging multiple access methods and the multiple services’ requirements for future home intelligent environments. This paper introduces the work in progress in constructing a stable and efficient Broadband Multiple Modes Residential Gateway (BMMRG) which supports multiple access interfaces, multiple services, IPv6, security, QoS and remotely web management. It is mainly based on an IXP425 network processor and a Linux kernel. We first present the hardware and software architectures of BMMRG, and then we introduce their in-detailed implementations. In the meantime, an intelligent home system is proposed based on BMMRG and household appliances equipped with wireless and ZigBee adapters. Finally, the effectiveness and feasibility of BMMRG is verified through testing.

Flexible job shop scheduling problem (FJSP) is an important extension of the classical job shop scheduling problem, where the same operation could be processed on more than one machine and has a processing time depending on the machine used. The objective is to minimize the makespan, i.e., the total duration of the schedule. In this article, we propose a multi-agent model based on the hybridization of the tabu search (TS) method and particle swarm optimization (PSO) in order to solve FJSP. Different techniques of diversification have also been explored in order to improve the performance of our model. Our approach has been tested on a set of benchmarks existing in the literature. The results obtained show that the hybridization of TS and PSO led to promising results.

This paper proposes a novel preprocessing method for the illumination-variant color face image. The proposed method aims to balance the luminosity and the color variation by color adjustment, bilateral filtering, and luminosity adjustment of sub image, which were required form input image by row-column transform and form which the output image was required by row-column inverse transform. The experimental results show that this preprocessing method can help to improve the segmentation precision and has good speed and robustness.

To solve the real-time problem of face detection, considering the realization bottleneck of AdaBoost pure software algorithm, FPGA-based hardware acceleration platform strategy is proposed. The paper analyzes the algorithm and partition the module for accelerating. The ZYNq-7000 platform FPGA from XILINX is adopted in the experiment, in which the hardware and software co-design is used. The final results show that it can detect face sat a 17fps speed with high hit rate and low false detection rate.

Remote sensing image segmentation algorithms are proposed for different thresholds with rough sets theory and fuzzy sets theory in this paper. The target and background fuzzy sets are gotten with the gray image as a fuzzy sets ; The target and background fuzzy sets are approximated by two rough fuzzy sets, the optimal image segmentation threshold is chosen by the optimal standard, Experimental results show that the proposed algorithms are more effective and flexible.

This paper describes a real-time noisy image edge detector to remove the noise which will bring negative effects on extraction and detection of image features. The average filtering algorithm is used to eliminate the noise of the original image and the Sobel edge detection operator is used to obtain image data. Both of the operation completes the functions including image acquisition and processing. Feasible verification of the edge detector is implemented in Altera EP3C55 using Verilog HDL language. Experimental results show that the edge detector is adaptive to the environment and can extract the noisy image edge effectively and promptly.

Pedestrian detection is widely used in automotive assisting driving system. The algorithm based on Histograms of Oriented Gradient (HOG for short) feature is the main one in the current pedestrian detection. This paper uses tri-linear interpolation method to extract the image HOG features, and gives the optimization algorithm based on look-up table to reduce the amount of calculation in extracting HOG feature. And then classifies them by RBF and linear SVM to explore its speed and accuracy. At the end of the paper, an effective method is given to merge windows that contain detected pedestrians. Experiments on INRIA and MIT databases show that the detecting accuracy and speed of this method is relatively high.

The visual system is one of the main equipment of unmanned surface vehicle (USV) autonomous navigation. Under the sea fog, atmospheric particles scattering leads to serious image degradation of the visual system. Because there is obvious sea-sky-line and the larger sky area in the image of offshore, so firstly, the image segmentation is done to get sky area, and through anglicizing sky area characteristics, the sky brightness is estimated, and then a simplified physical model of atmospheric scattering is built up, lastly image scene recovery is finished. Thinking about using this simple image defogging method to video image, foreground and background separation is done. Comparative research with several defogging methods onshore, results show that the proposed method can enhance the video image clarity of the USV visual system under sea fog very well. This research brought a good foundation to further improve the accuracy and precision of surface target identification and tracking algorithm.

In this paper, a real-time lane detection algorithm based on vision is presented. This algorithm improves the robustness and real-time of processing by combining with the dynamic region of interest (ROI) and the prior knowledge. When the lanes detected from previous frames have little changes for several frames, we recognize the lane only in dynamic ROI. We also proposed an erosion operator to refine the edge and a Hough transform with a restrict search space to detect lines with a faster rate. Experiments in structured road showed that the proposed lane detection method can work robustly in real-time, and can achieve a speed of 30ms/frame for 720×480 image size.

Nowadays a lot of information in the form of digital content is easily accessible but finding the relevant image is a big problem. This is where the Content Based Image Retrieval (CBIR) comes in to solve the image retrieval dilemma. But a CBIR system faces certain problems such as a strong signature development. Also, one of the major challenges of CBIR is to bridge the gap between the low level features and high level semantics. Previously, several researchers have proposed to improve the performance of a CBIR system but they have only answered image retrieval problem to an extent. In this paper, we propose a new CBIR signature that uses color color histogram. The results of the proposed method are compared previous method from the literature. The results of the proposed system demonstrates high accuracy rate than the previous systems in the simulations. The proposed system has significant performance.

This study investigates joint kinematics, joint angular positions, and orbital dynamic stability of human walking with different prosthetic feet by using graphical methods of phase plane portraits, Poincaré maps and Floquet multipliers, respectively. The Flex foot, SACH foot, Seattle foot and one non-specific optimized foot are taken as the research objects. Numerical experiments are performed to compare and evaluate human locomotion traits on several aspects by focusing on the concerned four kinds of prosthetic feet.

In this paper, we have proposed an improved Intelligent Water Drop (IWD) Algorithm. The IWD algorithm has been proposed by observing the dynamic flow of water in the river system and the actions of the water drops. The water drops act as agents to find the optimal solution. In this paper, we have modified the original IWD algorithm and proposed an improved variant of it. We have implemented our proposed algorithms to solve a real-life waste collection problem. Our algorithms have shown promising results.

Encoding method is one of the key factors of evolutionary design of analog circuit. Due to the adaptability, convenience and relatively short length of linear coding method, it has been widely used for automation of analog circuit design. Evolutionary design of analog circuits, which is not limited to traditional knowledge, could generate circuits with novel structures and parameters. The novel structures provide more possible solutions for fault-tolerance design of analog circuits. While, the current linear coding method based on five connection ways limits the number of possible circuit structures. So in this paper, we improve the existing linear coding method by expanding the instruction set. The experimental results show that the improved linear coding method can generate richer circuit structures, and it opens up a new way for the fault-tolerance design of analog circuits.

In recent years, the highly growth and development of world economy results in the natural resources being gradually run out and the environment further directly and indirectly being polluted more severe. Consequently, any kind of alternative energy resource have been developed, harvested and designed. An electric bicycle based on a blushless dc motor drive which has high efficiency, zero pollution, clean and convenient, is then designed and implemented in this paper. The hardware design based on a microcontroller is analyzed and discussed. The software programming is developed in MPLAB integrated development environment from the Microchip Technology Inc. Finally, a prototype of blushless dc motor drive for an electric bicycle is realized and demonstrated. The experimental results show the feasibility and fidelity of the complete designed system.

The Model Driven Architecture (MDA) methodology requires several intelligent operation stages, such as the computation independent model transformation (CIMT), the platform independent model transformation (PIMT), and the platform specific model transformation (PSMT), to progressively transform an abstract model to a physical system. The special Unified Modeling Language (UML) or StarUML is the core tool of CIMT that models a digital system in a diagram paradigm. PIMT uses the Python language with

minidom

object to perform a series translation from UML diagram to VHSIC Hardware Description Language (VHDL) code. Finally, the PSMT imports an

os

object to Python for running a series of synthesis command script to get bit stream that is finally downloaded into FPGA device to complete the realization of the digital logic circuit.

Traditional design flow is not efficient enough for the complex surface micromachined device because of its lack of perceptual intuition. The feature technology was introduced into micro device area to provide the comfortable design environment. The problem solving of generating mask from the 3D model of micro device plays the key role to implement the advanced design way, which emphasized on the structural design method instead of beginning with the issues of fabricating processes. In this paper, the algorithm to evaluate the data of mask based on the 3D model of surface micromachined device is presented. With respect to the etching process, the etched solids were built up by particular operations of the layer models to indicate the etched parts. After that, the mask is derived on the basis of the etched solids.

Virtual screening (VS) is a advanced technology to find some potential drugs from a large scale of small molecules, which will generate intensive results data. The data processing in VS is a key problem because of its scale. In the present work, we performed a study of exploiting MapReduce computing model to Parallel Process of Virtual Screening Result File. The results showed that it costs much less time using the combined methods. Hence, we recommend the efficient methods to process the results from large scale virtual screening.

Along with the development of fabricating processes, the structure of micro device becomes more and more complex. The traditional design tools begin with the processes design, which is not in a perceptually intuitive way. The 3D modeling method is presented to improve design efficiency. Above all, the data structure of the feature-based model is illustrated, by which the inner data of the 3D components to build up micro device is organized. Then, the 3D visualization environment is constructed to make the joint between the inner data and the interactive scene that receives the commands from the designers. By using this method, designer can build up the 3D model of micro device with the more convenient way.

Sebacate pillared layered double hydroxides (LDH) was prepared via co-precipitation method. And then the drug 10-hydroxy-camptothecin (10-HC) was intercalated into the gallery of LDH to form the drug–LDH composites. Three types of dissolution- diffusion kinetics models were used to make clear the drug release mechanism of the LDH composites. For the first time data reconciliation was used to make clear the drug release kinetics mechanism of the LDH composites, which was carried out by using a data filter algorithm based on first-order delay filter and time-delay principle to diminish random experimental errors resulting from the factors including stirring speed, solid content, and so on. Simulation results indicated that the data reconciliation adequately degraded the interactions between drug release rates at different times and made the data satisfy the kinetics models for the drug release mechanism more accurately.