Showing papers presented at "Intelligent Systems Design and Applications in 2006"

Robot Path Planning Based on Artificial Potential Field Approach with Simulated Annealing

Abstract: The artificial potential field (APF) approach provides a simple and effective motion planning method for practical purpose. However, artificial potential field approach has a major problem, which is that the robot is easy to be trapped at a local minimum before reaching its goal. The avoidance of local minimum has been an active research topic in path planning by potential field. In this paper, we introduce several methods to solve this problem, emphatically, introduce and evaluate the artificial potential field approach with simulated annealing (SA). As one of the powerful techniques for escaping local minimum, simulated annealing has been applied to local and global path planning.

Genetic Approach for Network Planning in the RFID Systems

Abstract: The problem of choosing the optimum locations for readers (antennas) in a RFID communications system is considered. All these choices must satisfy a set of imperative constraints and optimize a set of objectives. The factors affecting optimum selection are the complex propagation environments, the undesired mutual coverage and the unavoidable interference of multiple readers. Unlike the antenna positioning in traditional cellular networks, uplink signals, i.e. signals from tag towards reader, must be taken into account when dealing with the planning in the RFID networks. This paper presents a genetic approach for tackling this complex optimization problem. To validate this approach, computational results are presented for a typical test scenario.

Shadow Removal from a Single Image

Abstract: Shadow detection and removal in real scene images is always a challenging but yet intriguing problem. In contrast with the rapidly expanding and continuous interests on this area, it is always hard to provide a robust system to eliminate shadows in static images. This paper aimed to give a comprehensive method to remove both vague and hard shadows from a single image. First, classification is applied to the derivatives of the input image to separate the vague shadows. Then, color invariant is exploited to distinguish the hard shadow edges from the material edges. Next, we derive the illumination image via solving the standard Poisson equation. Finally, we got the shadow-free reflectance image. Experimental results showed that our method can robustly remove both vague and hard shadows appearing in the real scene images.

An Improved Genetic Algorithm of Optimum Path Planning for Mobile Robots

Abstract: An improved genetic algorithm of optimum path planning for mobile robots is proposed in this paper. An obstacle avoidance algorithm is introduced to generate the initial population in order to improve the path planning efficiency. Domain heuristic knowledge based crossover, mutation, refinement and deletion operators are specifically designed to fit path planning for mobile robots. Furthermore, a fuzzy logic control algorithm is integrated to self-adaptively adjust the probabilities of crossover and mutation in the genetic algorithm. Simulation studies for both static and dynamic environments are carried out, and the simulation results show that the proposed genetic algorithm exhibits improved search speed, high search quality and enhanced self-adaptability.

An Approach to Spam Detection by Naive Bayes Ensemble Based on Decision Induction

Abstract: Spam has been a serious problem to global email users. In this paper, a two-layered spam detection flow was used, which showed the trade-off between accuracy and efficiency. Then we discussed Naive Bayes classifiers ensemble based on Bagging. By casting spam detection in a decision theoretic framework, a Naive Bayes Bagging spam detection model based on embedded decision tree is proposed. Then this model was reduced by strict likelihood score bound limitation of the Naive Bayes classifiers. Finally, an improved method based on classifier error weighted is presented. The experiment results show that the modification is effective.

Recognition of ECG Patterns Using Artificial Neural Network

Abstract: In this paper, the artificial neural network method was used for Electrocardiogram (ECG) pattern recognition. Four types of ECG patterns were chosen from the MIT-BIH database to be recognized, including normal sinus rhythm, premature ventricular contraction, atrial premature beat and left bundle branch block beat. ECG morphology and R-R interval features were performed as the characteristic representation of the original ECG signals to be fed into the neural network models. Three types of artificial neural network models, SOM, BP and LVQ networks were separately trained and tested for ECG pattern recognition and the experimental results of the different models have been compared. The SOM network exhibited the best performance and reached an overall accuracy of 95.5%, and the BP and LVQ network reached 92.5% and 91.5%.

Geometrical Algorithms of Ear Contour Shape Representation and Feature Extraction

Abstract: In the article we present discrete geometrical algorithms of open contours representation, feature extraction and recognition. The proposed algorithms were developed for ear biometrics, but they can be also applied in other contour image processing applications. Firstly we present mathematical and algorithmic foundations of our geometrical feature extraction methods. We also discuss their application and the achieved results for ear contour image analysis and recognition.

A New Method to Assist Small Data Set Neural Network Learning

Abstract: Artificial neural networks are relevant to solve large sample problems and the learning performance may not be good in small sample conditions. Inspired by applications of posterior probability, a new neural network learning method based on posterior probability (PPNN) is proposed to improve small data set learning accuracy in this paper. Together with the techniques of creating new learning samples to fill up the gaps between original samples and using support vector machine (SVM) to obtain posterior probabilities, a novel neural network model whose inputs include the samples and their posterior probabilities is constructed. Simulation experiment and two real data application results indicate that learning accuracy can be significantly improved by the proposed algorithm involving very small data set. It provides a new feasible way to assist small data set neural network learning.

An Improved Learning Algorithm Based on The Broyden-Fletcher-Goldfarb-Shanno (BFGS) Method For Back Propagation Neural Networks

Abstract: The Broyden-Fletcher-Goldfarb-Shanno (BFGS) optimization algorithm usually used for nonlinear least squares is presented and is combined with the modified back propagation algorithm yielding a new fast training multilayer perceptron (MLP) algorithm (BFGS/AG). The approaches presented in the paper consist of three steps: (1) Modification on standard back propagation algorithm by introducing "gain variation" term of the activation function, (2) Calculating the gradient descent on error with respect to the weights and gains values and (3) the determination of the new search direction by exploiting the information calculated by gradient descent in step (2) as well as the previous search direction. The new approach improved the training efficiency of back propagation algorithm by adaptively modifying the initial search direction. Performance of the proposed method is demonstrated by comparing to the Broyden-Fletcher-Goldfarb-Shanno algorithm from neural network toolbox for the chosen benchmark. The results show that the number of iterations required by this algorithm to converge is less than 15% of what is required by the standard BFGS and neural network toolbox algorithm. It considerably improves the convergence rate significantly faster because of it new efficient search direction.

Correlation-based Feature Selection Strategy in Neural Classification

Abstract: One of the problems that have to be overcome in classification tasks is high data dimensionality. Therefore, dimensionality reduction techniques such as feature selection have to be employed. Feature selection involves univariate or multivariate evaluation of features with respect to the classification accuracy. Pairwise feature selection was recently proposed as a trade-off between selection process complexity and the need to analyze relationships between features. In our previous work we have proposed a correlation-based modification of the pairwise feature selection. In this paper we present the results of the experiments in which we have compared the correlation-based feature selection strategy with the unmodified pairwise approach. The experiments were performed using neural network classifiers on commonly used benchmark data sets.

Study and Simulation of Based-fuzzy-logic Parallel Hybrid Electric Vehicles Control Strategy

Abstract: With the increasing demand of fuel and emission Hybrid Electric Vehicles (HEV) with dual power sources, the engine and the motor, has been one of the development ways of clean automobiles. In the research on Hybrid Electric Vehicles, the control of the powertrain is the key issue. According to Parallel Hybrid Electric Vehicles, this paper used fuzzy control to realize Electric Assist Control Strategy and Membership Function was optimized for increasing fuel economy and decreasing emissions.

Anomaly Intrusion Detection Approach Using Hybrid MLP/CNN Neural Network

Abstract: An anomaly intrusion detection approach based on Hybrid MLP/CNN (Multi-layer Perceptron / Chaotic Neural Network) neural network is proposed in this paper. Most anomaly detection approaches using MLP can detecting novel real-time attacks, but still have high false alarm rates. Most attacks are composed of a series of anomaly events. These attacks are called timedelayed attacks, which current neural network IDSs (Intrusion Detection System) cannot identify efficiently. A hybrid MLP/CNN neural network is constructed in order to improve the detection rate of time-delayed attacks. While obtaining a similarly detection rate of real-time attacks as the MLP does, the proposed approach can detect time-delayed attacks efficiently with chaotic neuron. This approach also exhibits a lower false alarm rate when detects novel attacks. The simulation tests are conducted using DARPA 1998 dataset. The experimental results are presented and compared in ROC curves, which can demonstrate that the proposed approach performs exceptionally in terms of both detection rate and false alarm rate.

Improved Ant Colony Algorithm and its Applications in TSP

Abstract: In the fields of ant colony optimization (ACO), models of collective intelligence of ants are transformed into useful optimization techniques. A kind of improved ACO (named PMACO) approach for traveling salesman problems (TSP) is presented. Aimed at the disadvantages existed in ACO, several new betterments are proposed and evaluated. In particular, the option that an ant hunts for the next step, the use of a combination of two kinds of pheromone evaluation models, the change of amount in the ant colony during the run of the algorithm, and the mutation of pheromone are studied. We tested ACO algorithm on a set of benchmark problems from the Traveling Salesman Problem Library. It performed better than the original and the other improved ACO algorithms.

A Hybrid HMM-SVM Method for Online Handwriting Symbol Recognition

Abstract: This paper presents a combined approach for online handwriting symbols recognition. The basic idea of this approach is to employ a set of left-right HMMs as a feature extractor to produce HMM features, and combine them with global features into a new feature vector as input, and then use SVM as a classifier to finally identify unknown symbols. The new feature vector consists of the global features and several pairs of maximum probabilities with their associated different model labels. A recogniser based on this method inherits the practical and dynamical modeling abilities from HMM, and robust discriminating ability from SVM for classification tasks. This technique also reduces the dimensions of feature vectors significantly and solves the speed and size problem when using only SVM. The experimental results show that this combined hybrid approach outperforms several classifiers reported in recent researches, and could achieve recognition rates of 97.48%, 91.99% and 91.74% for digits and upper/lower case characters respectively on the UNIPEN database benchmarks.

Automatic Recognition Algorithm of Quick Response Code Based on Embedded System

Abstract: The automatic recognition algorithm of Quick Response Code is discussed in this paper. An image processing system based on embedded system is described to be able to binarization, location, segment, and decoding the QR Code. In order to adapting various sizes, various gray-level values, and under various lighting conditions of real bar code image, a high-speed, high-accuracy binarization method is developed, which can locate the finder pattern accurately and integrate the local thresholding method with global thresholding. Experiments have shown that over 99% barcode can be optimally recognized with the proposed algorithm. It can achieve higher recognition rate of high density bar code, and is applicable to real world scene image.

A Remote User Authentication Scheme Using Bilinear Pairings and ECC

Abstract: Recently, a novel remote user authentication scheme using bilinear pairings was proposed. The scheme is insecure, however, against forgery attack and replay attack. In this paper we propose a new design for remote user authentication using the properties of bilinear pairings, elliptic curve ElGamal encryption and decryption scheme. By employing EC ElGamal encryption and decryption, the replay attacks and the forgery attacks can be blocked. Moreover, a new strong password change method, which overcomes the password changing weakness, is provided. Finally we analyze the security and performance of our new design.

Task Scheduling Based on PSO Algorithm in Computational Grid

Abstract: Task scheduling is a key problem concerned in computational grid. In this paper, a heuristic approach based on particle swarm optimization is adopted to solving scheduling problem in grid environment. Each particle is represented a possible solution, and the position vector is transformed from the continuous variable to the discrete variable. This approach aims to generate an optimal schedule so as to get the minimum makespan and maximum resource utilization while completing the tasks. The results of simulated experiments show that the particle swarm optimization algorithm is able to get the better schedule than genetic algorithm.

Spatial Information Based Image Segmentation Using a Modified Particle Swarm Optimization Algorithm

Abstract: This article proposes a particle swarm based segmentation algorithm for automatically grouping the pixels of an image into different homogeneous regions. In contrast to most of the existing evolutionary image segmentation techniques, we have incorporated spatial information into the membership function for clustering. The spatial function is the summation of the membership function in the neighborhood of each pixel under consideration. The two very important advantages of the new method are: 1) It does not require a priori knowledge of the number of partitions in the image and 2) It yields regions, more homogeneous than the existing methods even in presence of noise.

JADE-Based A-Team as a Tool for Implementing Population-Based Algorithms

Abstract: The paper proposes a JADE-based A-Team environment as a middleware supporting the implementation and execution of population-based algorithms. The paper includes an overview of the JADE-based A-Team components and presents examples of the population-based architectures designed to obtain solutions to example combinatorial optimization problems including traveling salesman, resource-constrained project scheduling, vehicle routing and clustering problems. Conclusions focus on advantages of the JADE-based A-Team and on suggestions for further research

Solving the App-Level Classification Problem of P2P Traffic Via Optimized Support Vector Machines

Abstract: Since the emergence of peer-to-peer (P2P) networking in the last 90s, P2P traffic has become one of the most significant portion of the network traffic. Accurate identification of P2P traffic makes great sense for efficient network management and reasonable utility of network resources. App-level classification of P2P traffic, especially without payload feature detection, is still a challenging problem. This paper proposes a new method for P2P traffic identification and app-level classification, which merely uses transport layer information. The method uses optimized Support Vector Machines to perform large learning tasks, which is common in network traffic identification. The experimental results show that the proposed method has high efficiency and promising accuracy.

A Novel Spatial Clustering with Obstacles Constraints Based on Genetic Algorithms and K-Medoids

Abstract: Spatial clustering is an important research topic in spatial data mining (SDM). Many methods have been proposed in the literature, but few of them have taken into account constraints that may be present in the data or constraints on the clustering. These constraints have significant influence on the results of the clustering process of large spatial data. In this paper, we discuss the problem of spatial clustering with obstacles constraints and propose a novel spatial clustering method based on genetic algorithms (GAs) and K-Medoids, called GKSCOC, which aims to cluster spatial data with obstacles constraints. It can not only give attention to higher local constringency speed and stronger global optimum search, but also get down to the obstacles constraints and practicalities of spatial clustering. The results on real datasets show that it is better than standard GAs and K-Medoids

A Design Tool to Reason about Ambient Assisted Living Systems

Abstract: This paper proposes a design tool to investigate the properties and emergent behaviours of a special class of Ambient Assisted Living systems, namely mutual assistance communities where the dwellers contribute to each other?s well being. Purpose of our system is to understand how mutual assistance communities work, what consequences a design decision could ultimately bring about, and how to construct care communities providing timely and cost-effective service for elderly and disabled people. We prove that mutual assistance between dwellers can provide care in time, and decrease the requirement for professional medical service. The simulation results show that with the existing rules most of the requirements for help can be solved or promptly initiated inside the community before their members resort to external professionals.

Simulation on a Fuzzy-PID Position Controller of the CNC Servo System

Abstract: Variation of the system parameters and external disturbances always happen in the CNC servo system. With a traditional PID controller, it will cause large overshoot or poor stability. In this paper, a fuzzy-PID controller is proposed in order to improve the performance of the servo system. The proposed controller incorporates the advantages of PID control which can eliminate the steady-state error, and the advantages of fuzzy logic such as simple design, no need of an accurate mathematical model and some adaptability to nonlinearity and time-variation. The Fuzzy-PID controller accepts the error( e ) and error change( ec ) as inputs ,while the parameters kp , ki , kd as outputs. Control rules of the controller are established based on experience so that self-regulation of the values of PID parameters is achieved. A simulation model of position servo system is constructed in Matlab/Simulink module based on a high-speed milling machine researched in our institute. By comparing the traditional PID controller and the fuzzy-PID controller, the simulation results show that the system has stronger robustness and disturbance rejection capability with the latter controller which can meet the performance requirements of the CNC position servo system better.

Power Quality Disturbances Classification using Wavelet and Support Vector Machines

Abstract: Based on wavelet multiresolution analysis (MRA) and support vector machines (SVMs), a classification method for power quality disturbances in electrical power system is presented. After multiresolution signal decomposition of power quality disturbances, characteristic vectors can be obtained. Short time power transform (STPT) is also used to supplement the characteristic vectors from MRA. Support vector machines are used to classify these characteristic vectors of power quality disturbances, and the performance of SVMs is compared with that of artificial neural network (ANN).

The Particle Swarm: Individual and Collective Intelligence

Abstract: Western psychology has traditionally focused on processes considered to be internal or private to the individual, with the social world generally regarded as an aspect of "the environment." Recent cross-cultural psychological research reveals fundamental differences in the way cognition operates in people from different cultures, demonstrating that the social environment not only affects thought, but helps create it. These discoveries are mirrored in the field of computational intelligence; researchers identifying methods for eliciting intelligent behavior from machines are looking more and more into models that consider the individual inextricably integrated with the social milieu. These new models are radically different from traditional AI, which treats cognition as a set of processes taking place inside an isolated brain. In this lecture I will discuss these dichotomies in terms of the particle swarm algorithm, which is a model of collectively integrated intelligences; developments in the particle swarm paradigm will be framed in terms of the interplay of culture and cognition.

Particle Swarms as Video Sequence Inhabitants For Object Tracking in Computer Vision

Abstract: A solution based on a Swarm Intelligence metaphor with a prey-predator scheme is proposed for real time object tracking in video sequences, which is a basic process in multiple computer vision tasks. Swarm predator particles fly on a Boid-like fashion over image prey pixels, using combined image features to guide individual movement rules. Object tracking emerges from interaction between predator particles and their environment. The paper includes method?s description and experimental evaluations on video streams that illustrate the efficiency of swarm based methods in different vision tasks.

Application of Genetic Algorithm in Optimization of Fuzzy Control Rules

Abstract: The property of fuzzy controller is determined by the fuzzy-control rules. Optimization of fuzzy-control rules has been the bottle-neck problem of the fuzzy controller design. Based on thorough understanding of integrality, consistency and optimization of fuzzy-control rules, this paper provides an improved genetic algorithm for optimization of fuzzy-control rules, and also, it offers the operator of a genetic algorithm. The fuzzy controller designed with the optimized rules is verified by its simulation experiments.

Pipeline Damage and Leak Detection Based on Sound Spectrum LPCC and HMM

Abstract: In order to protect pipeline transportation and prevent from leakage incident for manmade damage or natural factors, it is very important to carry out such researches as active protecting and accurate positioning. Designed the pipeline prevention monitoring and leak detecting system based on calculating LPCC (Linear Prediction Cepstrum Coefficient) and using HMM (Hidden Markov Models) to recognise damage acoustic signals. The continuous non-steady time-variety process was sub-framed and described with a series of short steady sequences on the basis of acoustic signal characteristic analysed. LPCC which represents accurately each short-time acoustic signal was selected as the acoustic signal characteristic parameters and extracted effectively using Durbin algorithm; HMM was established to recognise damage types by Baum-Welch revaluation algorithm with the state-transfer probability and observing time sequences characteristic parameters; using Viterbi decoding algorithm realized the search of best transfer route and achieved the corresponding export probability. The results show that the acoustic singles recognition rate is improved effectively based on sound spectrum LPCC and HMM,and can be up to 97%.

Design Study of Adaptive Fuzzy PID Controller for LOS Stabilized System

Abstract: The LOS stabilization control based on gyro stabilized platform is required to isolate the line of sight (LOS) from the movement and vibration of carrier and ensure pointing and tracking for target in electro-optical tracking system. A hybrid adaptive fuzzy PID control is developed for this servo system with nonlinear property and some uncertainties. The self-tuning factors are made to modify the parameters of fuzzy controller online. A new learning algorithm of rules modifier is proposed to adjust control efforts. To eliminate the static error that fuzzy controller inheres and fulfil non-error control, the hybrid control are designed to achieve requirement for real-time and high stabilization precision. The experiment results in four-axes servo turntable are presented to verify the effectiveness of the proposed method in rejecting carrier disturbances

Graph-based Semi-supervised Learning Algorithm for Web Page Classification

Abstract: Many application domains such as web page classification suffer from not having enough labeled training examples for learning. However, unlabeled training examples are readily available but labeled ones are fairly expensive to obtain. As a result, there has been a great deal of work in resent years on semi-supervised learning. This paper proposes a graph-based semi-supervised learning algorithm that is applied to the web page classification. Our algorithm uses a similarity measure between web pages to construct a K-Nearest Neighbor graph. Labeled and unlabeled web pages are represented as nodes in the weighted graph, with edge weights encoding the similarity between the web pages. In order to use unlabeled data to help classification and get higher accuracy, edge weights of the graph are computed through combining weighting schemes and link information of web pages. The learning problem is then formulated in terms of label propagation in the graph. By using probabilistic matrix methods and belief propagation, the labeled nodes push out labels through unlabeled nodes. Our preliminary experiments on the WebKB dataset show that the algorithm in this paper can effectively exploit unlabeled data in addition to labeled ones to get higher accuracy of web page classification.