Search

In this paper, a digital image encryption scheme based on the hybrid chaotic system is proposed. A three-order Cellular Neural Network (CNN) with complex dynamical behavior and Logistic chaotic map are employed in the encryption scheme. The output sequences of two chaotic systems are combined effectively by using a special method, and a fast image encryption is realized by means of the bit-wise XOR operations. The results of the security analyses indicate that this encryption scheme not only has a large key space but also has a very sensitivity with respect to the encryption key. The encryption scheme proposed in this paper possesses perfect diffusion and confusion properties and it can resist the know-plaintext attacks and chosen-plaintext attacks. [ABSTRACT FROM AUTHOR]

This paper studied a model that is able to extract occluding surfaces of subjective contour figures based on the mechanism of feature extraction found in a visual system. A common factor in all such subjective contour figures, such as the Kanizsa triangle is having a surface occluding part of a background, i.e. subjective contours are always accompanied by subjective surfaces. In this paper we propose a neural network model that predicts the shape of subjective surfaces. This model employed an important two-stage process of the Induced Stimuli Extraction System (ISES) and Subjective Surfaces Perception System (SSPS). The former system extracted the induced stimuli for the perception of subjective surfaces, and the latter formed the subjective surfaces from the induced stimuli. The proposed model is demonstrated on a variety of Kanizsa-type subjective contour displays. The results of the experiment showed that the proposed model was successful not only in extracting the induced stimuli for the perception of subjective contours, but also in perceiving the subjective surface from the induced stimuli. [ABSTRACT FROM AUTHOR]

In this paper, a new filter, η-LM, which is based on Levenberg-Marquardt Artificial Neural Networks, is proposed for the impulsive noise suppression from highly distorted images. The η-LM uses Anderson-Darling goodness-of-fit test in order to find corrupted pixels more accurately. The extensive simulation results show that the proposed filter achieves a superior performance to the other filters mentioned in this paper in the cases of being effective in detail preservation and noise suppression, especially when the noise density is very high. [ABSTRACT FROM AUTHOR]

A new method is brought forward to predict multivariate time series in this paper. Related time series instead of a single time series are applied to obtain more information about the input signal. The input data are embedded as the phase space points. By the Principle Component Analysis (PCA) the most useful information is extracted form the input signal and the embedding dimension of the phase space is reduced, consequently, the input of the neural networks is simplified. The recurrent neural network has a number of advantages for predicting nonlinear time series. Therefore, Elman neural network is adopted to predict multivariate time series in this paper. Simulations of nonlinear multivariate time series from nature and industry process show the validity of the method proposed. [ABSTRACT FROM AUTHOR]

The focus of this paper is to propose an explanation for how biological auditory mechanism is able to use spiking neurons to code high bandwidth information using information channels with very slow sampling rates (< 20 Hz). The general approach described in this paper is to decompose the signal into narrow band channels, each of which can be sampled at a frequency that is much lower than the center frequency of the corresponding narrow band filter. The new idea here is that the system can use non-uniform sampling to capture both the amplitude of the modulation and the phase of the carrier signal. In this paper, we first describe a system based on FFT analysis combined with overlap-add and a sampling process where magnitude is digitized but phase is represented using a temporal code of spiking neurons. The coding/decoding mechanism is based on the properties of the refractory period. We demonstrate that it is possible to reduce the bit rate to 50% by coding the carrier phase using the timing of the pulses. In the second part of this paper we show how a biological system may approximate the broadband auditory signal using spiking neurons in conjunction with a simple model of neural refractory period. [ABSTRACT FROM AUTHOR]

This paper discusses blind identification and deconvolution of two-input two-output channels corrupted by noises based on second-order statistics. First, the identifiability of channel is analyzed. By constructing an new criterion, the channel parameters can be identified precisely in the present of noises. Second, the cost function of identification is established and the corresponding algorithm is presented. Next, a feedback model is used for deconvolution, and several important problems, such as the effect of noises in the blind deconvolution of mixed sources and the stability of deconvolution model, are discussed. At last, simulation results are given to illustrate the theoretical results of this paper. [ABSTRACT FROM AUTHOR]

This paper presents a method of processing Chinese syntactic category ambiguity with support vector machines (SVMs): extracting the word itself, candidate part-of-speech (POS) tags, the pair of candidate POS tags and their probability and context information as the features of the word vector. A training set is established. The machine learning models of disambiguation based on support vector machines are obtained using polynomial kernel functions. The testing results show that this method is efficient. The paper also gives the results obtained with neural networks for comparison. [ABSTRACT FROM AUTHOR]

This paper introduced a technique for authenticating the vehicle engines by comparing the images of the imprints of the identification number acquired when the vehicle was first registered and the ones acquired from the routine yearly vehicle inspection. The images are taken by rubbing a pencil over a piece of paper covered over the images and then are scanned into a computer. Due to the nature of the acquiring technique, the acquired images have lots of artifacts caused by the shape and the condition of the engine surface and unevenness of rubbing the pencils by hand. We used the polynomial spline pyramid algorithm to acquire a training set using ART-2, which is considered a tradeoff of stability-plasticity dilemma. The experiments show an accuracy rate close to 80%. [ABSTRACT FROM AUTHOR]

This paper proposed the intelligent immigration control system that authorizes the traveler immigration and detects the forged passports by using automatic recognition of passport codes and the picture and face verification. The proposed system extracts and deskewes the areas of passport codes from the passport image. This paper proposed a novel ART algorithm creating the adaptive clusters to the variations of input patterns and it was applied to the extracted code areas for the code recognition. After compensating heuristically the recognition result, the detection of forged passport is achieved by using the picture and face verification between the picture extracted from passport image and the picture retrieved from the database based on the recognized codes. Due to the proposed ART algorithm and the heuristic refinement, the proposed system shows the relatively better performance. [ABSTRACT FROM AUTHOR]

A novel clustering algorithm based upon a SOFM neural network family is proposed in this paper. The algorithm takes full advantage of the characteristics of SOFM Neural Network family and defines a novel similarity measure, topological similarity, which help the clustering algorithm to handle the clusters with arbitrary shapes and avoid suffering from the limitations of the conventional clustering algorithms. The paper suggests another novel thought to tackle the clustering problem. [ABSTRACT FROM AUTHOR]

In this paper, Independent Component Analysis (ICA) is presented as an efficient face feature extraction method. In a task such as face recognition, important information may be contained in the high-order relationship among pixels. ICA is sensitive to high-order statistic in the data and finds not-necessarily orthogonal bases, so it may better identify and reconstruct high-dimensional face image data than Principle Component Analysis (PCA). ICA algorithms are time-consuming and sometimes converge difficultly. A modified FastICA algorithm is developed in this paper, which only need to compute Jacobian Matrix one time in once iteration and achieves the corresponding effect of FastICA. Finally a genetic algorithm is introduced to select optimal independent components (ICs). The experiment results show that modified FastICA algorithm quickens convergence and genetic algorithm optimizes recognition performance. ICA based features extraction method is robust to variations and promising for face recognition. [ABSTRACT FROM AUTHOR]

This paper proposes a Foremost-Policy Reinforcement Learning based ART2 neural network (FPRL-ART2) and its learning algorithm. For real time learning, we select the first awarded behavior based on current state in the Foremost-Policy Reinforcement Learning (FPRL) in stead of the optimal behavior in 1-step Q-Learning. The paper also gives the algorithm of FPRL and integrates it with ART2 neural network. ART2 is used for storing the classified pattern and the stored weights of classified pattern is increased or decreased by reinforcement learning. FPRL-ART2 is successfully used in collision avoidance of mobile robot and the simulation experiment indicates that collision times between robot and obstacle are decreased effectively. FPRL-ART2 makes favorable effect against collision avoidance. [ABSTRACT FROM AUTHOR]

This paper is concerned with the problem of Reinforcement Learning (RL) in large or continuous spaces. Function approximation is the main method to solve such kind of problem. We propose using neural networks as function approximators in this paper. Then we experiment with three kind of neural networks in Mountain-Car task and illustrate comparisons among them. The result shows that CMAC and Fuzzy ARTMAP perform better than BP in Reinforcement Learning with Function Approximation (RLFA). [ABSTRACT FROM AUTHOR]

The extension neural network types 1 and 2 have been proposed in my recent paper. In this sequel, this paper will propose a new extension neural network called ENN-3. The ENN-3 is a three layers neural network and a pattern classifier. It shows the same capability as human memory systems to keep stability and plasticity characteristics at the same time, and it can produce meaningful weights after learning. The ENN-3 can solve the linear and nonlinear separable problems. Experimental results from two benchmark data sets verify the effectiveness of the proposed ENN-3. [ABSTRACT FROM AUTHOR]

This paper studies the strength and direction of phase synchronization among Neural Networks (NNs). First, a nonlinear lumped-parameter cerebral cortex model is addressed and used to generate epileptic surrogate EEG signals. Second, a method that can be used to calculate the strength and direction of phase synchronization among NNs is described including the phase estimation, synchronization index and phase coupling direction. Finally, simulation results show the method addressed in this paper can be used to estimate the phase coupling direction among NNs. [ABSTRACT FROM AUTHOR]

In this paper, we have obtained some sufficient conditions to guarantee that Cohen-Grossberg neural networks with discrete and distributed delays have a periodic orbit and this periodic orbit are globally attractive. The results presented in this paper are the improvement and extension of the existed ones in some existing works. Finally, the validity and performance of the results are illustrated by two simulation examples. [ABSTRACT FROM AUTHOR]

This paper discusses impulsive control and synchronization of interval Hopfield neural networks (HNN for short). Based on the matrix measure and new comparison theorem, this paper presents an impulsive robust control scheme of the interval HNN. We derive some sufficient conditions for the stabilization and synchronization of interval Hopfield neural networks via impulsive control with varying impulsive intervals. Moreover, the large upper bound of impulsive intervals for the stabilization and synchronization of interval HNN can be obtained. [ABSTRACT FROM AUTHOR]

Recently, there are several papers discussing global robust stability of the equilibrium point for the interval delayed neural networks. However, we find these criteria are not accurate. In this paper, based on Linear Matrix Inequality (LMI) technique, we propose an algorithm to determine in which region the interval delayed system is globally robust stabile. This approach is much more powerful than the criteria given in previous papers. We also give a numerical example to illustrate the viability of our algorithm. [ABSTRACT FROM AUTHOR]

The paper studies the nonautonomous delayed recurrent neural networks. By applying Lyapunov functional method and utilizing the technique of inequality analysis, we obtain the sufficient condition to ensure the globally asymptotic stability and globally exponential stability. The results given in this paper are new and useful. [ABSTRACT FROM AUTHOR]

For the general multi-input and multi-output architecture of multilayer perceptrons, the issue of classes of congruent error hypersurfaces is converted into the issue of classes of congruent pattern sets. By finding the latter number which is much smaller than the total number of error hypersurfaces, the complexity of error hypersurfaces is reduced. This paper accomplishes the remaining work left by [4] which only addresses multi-input and single-output architecture. It shows that from the input side, group G(N) includes all the possible orthogonal operations which make the error hypersurfaces congruent. In addition, it extends the results from the case of single output to the case of multiple outputs by finding the group S(M) of orthogonal operations. Also, the paper shows that from the output side, group S(M) includes all the possible orthogonal operations which make the error hypersurfaces congruent. The results in this paper simplify the complexity of error hypersurfaces in multilayer perceptrons. [ABSTRACT FROM AUTHOR]

This paper addresses the problem of Concurrent Mapping and Localization(CML) by means of a hybrid immune evolutionary computation based on immunity and clonal selection for a mobile robot. An immune operator, a vaccination operator, is designed in the algorithm. The experiment results of a real mobile robot show that the computational expensiveness of the algorithm in this paper is less than other algorithms and the maps obtained are very accurate. [ABSTRACT FROM AUTHOR]

Conventionally the boilers in space heating systems are controlled by open-loop control systems due to the absence of a practical method for measuring the overall thermal comfort level in the building. This paper describes a neural-fuzzy based inferential sensor that can be used to design close-loop boiler control schemes. Both simulation and experimental results show that the proposed technique results in significant energy saving and improvement on the control of thermal comfort in the built environment. The paper also describes the ongoing and future work. [ABSTRACT FROM AUTHOR]

This paper presents a novel approach for visual inspection of textures. The approach applies the artificial immune theory to learning the filters for texture flaw detection, which are invariant to changes of texture orientations and scales. In this paper, defect textures and defect-free textures are regarded as non-self and self respectively, and texture filters are regarded as antibodies. The clonal selection based algorithm is presented to evolve antibodies. Experimental results on TILDA textile images were done to show the feasibility of the proposed method. [ABSTRACT FROM AUTHOR]

Most existing spreading models for network viruses are developed refereing to the epidemic models for biological viruses. However, Why most network viruses spread much slower than those models predicate? Why most network viruses still exist when they go beyond the threshold predicated by those models? Contrary to the prior models, the paper points out network viruses have different spreading features compared with biological viruses, such as the connectivity rate and cure rate are both functions of the time which are also key factors to affect the spreading of viruses. Based on which the paper constructs a more general epidemiological model for the network viruses. For several particular cases the paper presents the simulations of the connectivity rate and cure rate and find they are consistent well with the statistics of some real viruses. Thus the paper opens one path to modifying the traditional epidemic models. [ABSTRACT FROM AUTHOR]

This paper presents the rule generation method based on evolutionary computation and rough set, which integrates the procedure of discretization and reduction using information entropy-based uncertainty measures and evolutionary computation. Based on the definitions of certain rules and approximate certain rules, the paper focuses on the reduction by meanings of evolutionary computation. Experimental results reveal that the proposed method leads to better classification quality and smaller number of decision rules comparing with other methods. [ABSTRACT FROM AUTHOR]

The purpose of this paper is to introduce a recently-developed nature-inspired algorithm, harmony search (HS), and to apply the algorithm to water pump switching problem. The HS algorithm is conceptualized using the musical improvisation process of searching for a better state of harmony. This paper describes a HS algorithm-based approach for the optimal switching problem in serial water pumping system. A standard example from the literature is presented to demonstrate the effectiveness of the proposed method, and the results are compared to genetic algorithm and branch & bound method. Computational results indicate that the HS approach becomes a good optimization model for solving water pump switching problem. [ABSTRACT FROM AUTHOR]

This paper is mainly about methodology in designing quantum algorithm. Based on study of Grover's algorithm, we argue that it is a short cut to design and interpret quantum algorithms from the viewpoint of Householder transformation directly. We give an example for this claim, which extends Grover's quantum search algorithm to some structured database. In this example, we show how to exploit some special structure information of problem, which restricts the search in some subspace. Based on an instantiation of this framework, we show that it does can utilize the information to the full extent. This paper gives the details that produce the algorithm framework. The idea, which is simple and intelligible, is universal to some extent, and therefore can be applied to other similar situations. [ABSTRACT FROM AUTHOR]

Recently, many genetic algorithms (GAs) have been developed as an approximate method to generate Pareto frontier (the set of Pareto optimal solutions) to multi-objective optimization problem. In multi-objective GAs, there are two important problems : how to assign a fitness for each individual, and how to make the diversified individuals. In order to overcome those problems, this paper suggests a new multi-objective GA using generalized data envelopment analysis (GDEA). Through numerical examples, the paper shows that the proposed method using GDEA can generate well-distributed as well as well-approximated Pareto frontiers with less number of function evaluations. [ABSTRACT FROM AUTHOR]

The paper proposed a novel quantum-inspired genetic algorithm with only one chromosome, which we called Single-Chromosome Quantum Genetic algorithm (SCQGA). In SCQGA, by bringing the information representation in quantum computing into the algorithm, only one quantum chromosome (QC) is used to represent all possible states of the entire population. A novel quantum evolution method without using conventional genetic operators such as crossover operator and mutation operator is proposed, in which according to the best individuals generated by QC we adjust the quantum probability amplitude with quantum rotation gates so that the QC can produce more promising individuals with higher probability in the next generation. The paper indicated that SCQGA is a new approach belonging to estimation of distribution algorithms (EDAs). Experiments on solving a class of combinatorial optimization problems show that SCQGA performs better than conventional genetic algorithm. [ABSTRACT FROM AUTHOR]

This paper considers how to increase the capacities of the elements in a set E efficiently so that probability of the total cost for the increment of capacity can be under an upper limit to maximum extent while the final expansion capacity of a given family F of subsets of E is with a given limit bound. The paper supposes the cost w is a stochastic variable according to some distribution. Network bottleneck capacity expansion problem with stochastic cost is originally formulated as Dependent-chance programming model according to some criteria. For solving the stochastic model efficiently, network bottleneck capacity algorithm, stochastic simulation, neural network(NN) and genetic algorithm(GA) are integrated to produce a hybrid intelligent algorithm. Finally a numerical example is presented. [ABSTRACT FROM AUTHOR]

This paper presents a novel genetic algorithm (GA) for the scheduling problem of input-Queued switch, which can be applied in various networks besides the design of high speed routers. The scheduler should satisfy quality of service (QoS) constraints such as throughput and jitter. Solving the scheduling problem for the input-Queued switches can be divided into two steps: Firstly, decomposing the given rate matrix into a sum of permutation matrices with their corresponding weights; secondly, allocating the permutation matrices in one scheduling period based on their weights. It has been proved that scheduling problem in input-Queued switch with throughput and jitter constraints is NP-complete. The main contribution of this paper is a GA based algorithm to solve this NP-complete problem. We devise chromosome codes, fitness function, crossover and mutation operations for this specific problem. Experimental results show that our GA provides better performances in terms of throughput and jitter than a greedy heuristic. [ABSTRACT FROM AUTHOR]

In the research of MOEA (Multi-Objective Evolutionary Algorithm), many algorithms for multi-objective optimization have been proposed. Diversity of the solutions is an important measure, and it is also significant how to evaluate the diversity of an MOEA. In this paper, the clustering algorithm based on the distance between individuals is discussed, and a diversity metric based on clustering is also proposed. Applying this metric, we compare several popular multi-objective evolutionary algorithms. It is shown by experimental results that the method proposed in this paper performs well, especially helps to provide a comparative evaluation of two or more MOEAs. [ABSTRACT FROM AUTHOR]

This paper proposes an improved genetic algorithm for producing aesthetically pleasing drawings of general undirected graphs. Previous undirected graph drawing algorithms draw large cycles with no chords as concave polygons. In order to overcome such disadvantage, the genetic algorithm in this paper designs a new mutation operator single-vertex- neighborhood mutation and adds a component aiming at symmetric drawings to the fitness function, and it can draw such type graphs as convex polygons. The improved algorithm is of following advantages: The method is simple and it is easy to be implemented, and the drawings produced by the algorithm are beautiful, and also it is flexible in that the relative weights of the criteria can be altered. The experiment results show that the drawings of graphs produced by our algorithm are more beautiful than those produced by simple genetic algorithms, the original spring algorithm and the algorithm in bibliography [4]. [ABSTRACT FROM AUTHOR]

In the standard GA, the individual has no intelligence and must act upon some rules established by a programmer in advance, such as various genetic operator. The result is to make the evolutionary process to be trapped into the local optimization of the objective function. In order to solve this problem, through studying the structure of an agent and selection operator, the paper designs a new genetic algorithm based on agent, called AGA (Agent-based Genetic Algorithm). At the premise of giving the definition of the outer environment where an agent lives and of an agent's belief, this paper gives some rules on how an agent selects one agent to cross their genes and some rules on how to solve competition. In addition, a communication method based on blackboard is presented to solve the communication among the agent society. Finally, the paper gives the structure of AGA and the simulation result for a multi-peak function, which demonstrates the validity of the AGA. [ABSTRACT FROM AUTHOR]

This paper presents a new method for hierarchical image segmentation. The hierarchical structure is represented by a binary tree with the main image as its root. At the lower levels, each node stands as one image segment, which is described by a weighted graph and may be divided into two new segments at the next level through a specific cut. Graph bi-sectioning is done by the self organizing property of ant systems. Ants are free to wander over one image segment graph to find the best cut on it. When an ant finds a suitable cut, it returns to its colony and leaves a proper value of pheromone over its trail to attract other ants to that cut. By using the Chemical Computing approach in this paper, it is assumed the mobile hormones (pheromone) are secreted which can diffuse around initial positions and attract more ants to the found cut. The advantages of this assumption are reducing the noise effects and improving the convergence speed of ants to find a new selected image segment, which can be seen in the practical results. [ABSTRACT FROM AUTHOR]

Optimal design for urban mass transit network is the precondition and basis to establish an effective public transportation system. Transit network optimization and headway optimization are two of the most important issues to be dealt with. In this paper, a transit network optimization model is firstly proposed to maximize the nonstop passenger flow. Moreover, this paper puts forwards the optimization model of headways for all the transit routes in the optimized network. Since all the two models can be boiled down to the NP-hard problem, two kinds of evolutionary algorithms, i.e., ant colony algorithm and improved genetic algorithm are introduced to solve the problems respectively. Finally, a case study in a typical city is introduced to explain the validity of the proposed methods. [ABSTRACT FROM AUTHOR]

Memetic evolutionary algorithms (MEAs) combine the global search of evolutionary learning methods and the fine-tune ability of local search methods so that they are orders of magnitude more accurate than traditional evolutionary algorithms in many problem domains. However, little work has been done on the mathematical model and convergence analysis of MEAs. In this paper, a theoretical model as well as the convergence analysis of a class of gradient-based MEAs is presented. The results of this paper are extensions of the research work on the abstract model and convergence analysis of general evolutionary algorithms. By modeling the local search of gradient methods as an abstract strong evolution operator, the theoretical framework for abstract memetic evolutionary algorithms is derived. Moreover, the global convergence theorems and the convergence rate estimations of gradient-based MEAs are also established. [ABSTRACT FROM AUTHOR]

A new algorithm of evolutionary computing, which combines clone selective algorithm involved in artificial immunity system theory and mind evolution algorithm (MEA) proposed in reference [4], is presented in this paper. Based on similartaxis which is the one of MEA operators, some operators borne by the new algorithm including such as clone mutation, clone crossover, clone selection are also introduced. Then the clone mind evolution algorithm (CMEA) is developed by using the diversity principle of antigen-antibody. Not only can CMEA converge to globally optimal solution, but also it solve premature convergence problem efficiently. The simulating results of the representative evaluation function show that the problem of degeneration phenomenon existing in GA and MEA can be perfectly solved, and the rapidity of convergence is evidently improved by CMEA studied in the paper. In the example of the solution to the numerical problem, the search range of solution is expanded and the possibility of finding the optimal solution is increased. [ABSTRACT FROM AUTHOR]

According to the principles of non-self detection and negative selection in natural immune system, two generating algorithms of detector are proposed in this paper after reviewing current detector generating algorithms used in artificial immune systems. We call them as Bit Mutation Growth Detector Generating Algorithm (BMGDGA) and Arithmetical-compliment Growth Detector Generating Algorithm (AGDGA) based on their operational features. The principle and work procedure of the two detector generating algorithms are elaborated in details in the paper. For evaluation of the proposed algorithms, they are tested and verified by using different datasets, and compared to Exhaustive Detector Generating Algorithm (EDGA). It turns out that the proposed two algorithms are superior to EDGA in detection performance and computational complexities. [ABSTRACT FROM AUTHOR]

Artificial Immune Systems (AIS), which are inspired from nature immune system, have recently been investigated for many information processing applications, such as feature extraction, pattern recognition, machine learning and data mining. In this paper, we investigate AIS, and in particular the clonal selection algorithm for Associative Classification (AC). To implement associative classification effectively, we need to tackle the problems on the very large search space of candidate rules during the rule mining process. This paper proposes a new approach known as AIS-AC for mining association rules effectively for classification. In AIS-AC, we treat the rule mining process as an optimization problem of finding an optimal set of association rules according to some predefined constraints. The proposed AIS-AC approach is efficient in dealing with the complexity problem on the large search space of rules. It avoids searching greedily for all possible association rules, and is able to find an effective set of associative rules for classification. [ABSTRACT FROM AUTHOR]

Multiple classifier systems have shown a significant potential gain in comparison to the performance of an individual best classifier. In this paper, a weighted combination model of multiple classifier systems was presented, which took sum rule and majority vote as special cases. Particle swarm optimization (PSO), a new population-based evolutionary computation technique, was used to optimize the model. We referred the optimized model as PSO-WCM. An experimental investigation was performed on UCI data sets and encouraging results were obtained. PSO-WCM proposed in this paper is superior to other combination rules given larger data sets. It is also shown that rejection of weak classifier in the ensemble can improve classification performance further. [ABSTRACT FROM AUTHOR]

Topical crawlers are becoming important tools to support applications such as specialized Web portals, online searching, and competitive intelligence. A topic driven crawler chooses the best URLs to pursue during web crawling. It is difficult to evaluate what URLs downloaded are the best. This paper presents some important metrics and an evaluation function for ranking URLs about pages relevance. We also discuss an approach to evaluate the function based on GA. GA evolving process can discover the best combination of the metrics' weights. Avoiding misleading the result by a single topic, this paper presents a method which characterization of the metrics' combination be extracted by mining frequent patterns. Extracting features adopts a novel FP-tree structure and FP-growth mining method based on FP-tree without candidate generation. The experiment shows that the performance is exciting, especially about a popular topic. [ABSTRACT FROM AUTHOR]

Numerous studies on time series prediction have been undertaken by a lot of researchers. Most of them relate to the construction of structure-invariable system whose parameter values do not change all the time. In fact, the parameter values of many realistic systems are always changing with time. In this case, the embedding theorems are invalid, predicting the behavior of parameter-varying systems is more difficult. This paper presents a new prediction technique, which is multiwavelets neural network. This technique absorbs the advantage of high resolution of wavelets and the advantages of learning and feed-forward of neural networks. The procedure of using the multiwavelets neural network for predicting is described in detail in this paper. Principal components analysis (PCA) as a statistical technique has been used to simplify the time series analysis in our experiments. The effectiveness of this network is demonstrated by applying it to predict Ikeda time series. [ABSTRACT FROM AUTHOR]

The business field is one of the important fields where the data mining technology is applied. The study mainly focuses on different attribute object's quantitative prediction and customer structure's qualitative prediction. Aiming at the characteristics of time series in business field, such as near-periodicity, non-stationarity and nonlinearity, the wavelet-neural networks-ARMA method is proposed and its application is examined in this paper. The hidden period and the non-stationarity existed in time series are extracted and separated by wavelet transformation. The characteristic of wavelet decomposition series is applied to BP networks and an autoregressive moving average (ARMA) model. The given example elucidates that the forecasting method mentioned in this paper can be employed to the business field successfully and efficiently. [ABSTRACT FROM AUTHOR]

This paper investigates the similarity of two sequences, one of the main issues for fragments clustering and classification when sequencing the genomes of microbial communities directly sampled from natural environment. In this paper, we use the relative entropy as a criterion of similarity of two sequences and discuss its characteristics in DNA sequences. A method for evaluating the relative entropy is presented and applied to the comparison between two sequences. With combination of the relative entropy and the length of variables defined in this paper, the similarity of sequences is easily obtained. The SOM and PCA are applied to cluster subsequences from different genomes. Computer simulations verify that the method works well. [ABSTRACT FROM AUTHOR]

In this paper, based on the theory of nonseparable wavelet, a novel nonseparable wavelet model has been proposed. The structure of the model is distinguished from that of wavelet network (RBF structure). It is a four-layer structure, which helps overcome the structural redundancy. In the process of the training of the network, in the light of the characteristics of nonseparable wavelet, a novel method of setting the initial value of weight has been proposed. It can overcome the shortcoming of gradient descent methodology that it makes the convergence of the network slow. Some experiments with the novel model for function learning will be shown. Comparing with the present wavelet networks, BP network, the results in this paper show that the speed and generalization performance of the novel model have been greatly improved. [ABSTRACT FROM AUTHOR]

In this paper, we review a visual approach and propose it for analysing computer-network activity, which is based on the use of unsupervised connectionist neural network models and does not rely on any previous knowledge of the data being analysed. The presented Intrusion Detection System (IDS) is used as a method to investigate the traffic which travels along the analysed network, detecting SNMP (Simple Network Management Protocol) anomalous traffic patterns. In this paper we have focused our attention on the study of anomalous situations generated by a MIB (Management Information Base) information transfer. [ABSTRACT FROM AUTHOR]

Till now, the problem of neural coding remains a puzzle. The intrinsic information carried in irregular neural spikes sequence is not known yet. But solution of the problem will have direct influence on the study of neural information mechanism. In this paper, coding mechanism of the neural spike sequence, which is caused by input stimuli of various frequencies, is investigated based on analysis of H-H equation with the method of nonlinear dynamics. The signals of external stimuli - those continuously varying physical or chemical signals - are transformed into frequency signals of potential in many sense organs of biological system, and then the frequency signals are transformed into irregular neural coding. This paper analyzes in detail the neuron response of stimuli with various periods and finds the possible rule of coding. [ABSTRACT FROM AUTHOR]

In this paper we propose a new stochastic nonlinear evolution model that is used to describe activity of neuronal population, we obtain dynamic image of evolution on the average number density in three-dimensioned space along with time, which is used to describe neural synchronization motion. This paper takes into account not only the impact of noise in phase dynamics but also the impact of noise in amplitude dynamics. We analyze how the initial condition and intensity of noise impact on the dynamic evolution of neural coding when the neurons spontaneously interact. The numerical result indicates that the noise acting on the amplitude influences the width of number density distributing around the limit circle of amplitude and the peak value of average number density, but the change of noise intensity cannot make the amplitude to participate in the coding of neural population. The numerical results also indicate that noise acting on the amplitude does not affect phase dynamics. [ABSTRACT FROM AUTHOR]

This paper incorporates robustness into neural network modeling and proposes a novel two-phase robustness analysis approach for determining the optimal feedforward neural network (FNN) architecture in terms of Hellinger distance of probability density function (PDF) of error distribution. The proposed approach is illustrated with an example in this paper. [ABSTRACT FROM AUTHOR]