As cross-lingual information retrieval attracts increasing attentions, tools that measure semantic similarity between Chinese and English documents become desirable. Since the way that humans convey thoughts at the abstract concept level makes little, if any, difference in the languages they use, it is possible to measure semantic similarity between different lingual documents using their conveyed abstract concepts. One mechanism to represent concepts or knowledge is ontology, which specifies terms, their properties and relations. We believe ontology learning technologies are suitable to find the discourse structure of the text through building its content ontology. In this paper, a novel method for semantic similarity measure between Chinese and English documents by applying ontology learning technologies is proposed. The method builds and compares the content ontologies of documents for measuring their semantic similarity, in which WordNet is referenced for word sense disambiguation and semantic similarity computation. This method makes possible to distinguish among the documents whose original texts appear not similar but convey similar concepts. A prototype of this cross-lingual semantic similarity measure scheme is constructed and some Chinese-English parallel documents are sampled for evaluation. The result of applying this prototype to these documents is shown with discussions of possible further improvement.

View full abstract

Nowadays, Capability Maturity Model Integration (CMMI) is becoming more and more important in the software industry, but it often takes a large amount of effort, time and cost to introduce CMMI. This paper focuses on Requirements Management (REQM) process area of CMMI to develop a CMMI-supported tool with the Web services architecture. Owing to the fact that establishing and maintaining the bi-directional traceability among requirements are a time-consuming and complex task; therefore, in order to establish and maintain the bi-directional traceability among requirements more effectively, this paper utilizes ontology and fuzzy inference mechanism to develop a requirement trace service that can automatically generate horizontal and vertical traceability relationships. From the experimental results, this proposed Web services architecture can operate independently; meanwhile, it also can be extended to other process areas of CMMI.

View full abstract

Mining association rules from a large business database, has been recognized as an important topic in the data mining community. A method that can help the analysis of associations is the use of classification ontology (taxonomy) and the setting of parameter constraints, such as minimum support. In real world applications, however, the classification ontology cannot be kept static while new transactions are continuously added into the original database, and the analysts may also need to set a different support constraint from the original one while formulating a new query in discovering real informative rules. Efficiently updating the discovered generalized association rules to reflect the change with classification ontology, support constraint and new added transactions is a crucial task. In this paper, we examine this problem and propose a novel algorithm, called IMA_HOSU, which can incrementally update the discovered generalized association rules when the classification ontology updates with a renewed minimum support. Empirical evaluations show that our algorithm is faster than applying the contemporary generalized associations mining algorithms to the whole updated database.

View full abstract

Wireless networks and mobile applications have grown very rapidly and have made a significant impact on computer systems. Especially, the usage of mobile phones and PDA is increased very rapidly. Added functions and values with these devices are thus greatly developed. If some regularity can be known from the user mobility behavior, then these functions and values can be further expanded and used intelligently. This paper thus attempts to discover personal mobility patterns for helping systems provide personalized service in a wireless network. The classification and the duration of each location area visited by a mobile user are used as important attributes in representing the results. A data mining algorithm has then been proposed, which is based on the AprioriAll algorithm, but different from it in several ways. Experiments are also made to show the effect of the proposed algorithm.

View full abstract

In this paper, we propose a fast TD-error propagation method with considering the similarity between states. By defining the similarity between state, the propagation is taken along with not only the previous visited states, but also similar neighboring states. The parameters of the states, such as estimated values, the positions in the state space and action selection probability, are used for fuzzy clustering of the states. We verify the effectiveness of the proposed method with a computer experience of a simple maze problem.

View full abstract

In this paper, we propose a novel selection strategy for genetic algorithms (GA) using self-organizing maps (SOM) with two type input vectors. The type is decided according to whether an evaluation value of the input vector is positive or negative. The evaluations indicate valuable distribution of population for genetic process, and modulate a way of updating reference vectors of the SOM. The proposed selection strategy achieves maintenance of a genetic diversity by employing the SOM which is available for input vectors with evaluations. An effectiveness of the proposed selection strategy is verified by some benchmark simulations for GA.

View full abstract

In this paper, we propose an effective skew estimation algorithm using a half-cosine function wavelet network. The half-cosine function wavelet network uses a half of cosine as a basis function and is a kind of functional-link networks, which has a good performance of human-like feature extractions. Our previous work has shown that the headline areas on text documents can be extracted effectively by using the network. In this paper, we try to estimate the object skew from the center coordinates of the basis functions forming the extracted object image. The proposed method can execute the object area extraction and skew estimation simultaneously. Here, the validity of the proposed method is confirmed for estimation of character sequence's skew and the usability is discussed.

View full abstract

Autonomous Underwater Vehicles (AUVs) are attractive tools to survey earth science and oceanography, however, there exists a lot of problems to be solved such as motion control, acquisition of sensor data, decision-making, navigation without collision, self-localization and so on. In order to realize useful and practical robots, underwater vehicles should take their action by judging the changing condition from their own sensors and actuators, and are desirable to make their behavior, because of features caused by the working environment. We have been investigated the application of brain-inspired technologies such as Neural Networks (NNs) and Self-Organizing Map (SOM) into AUVs. A new controller system for AUVs using Modular Network SOM (mnSOM) proposed by Tokunaga et al. is discussed in this paper. The proposed system is developed using recurrent NN type mnSOM. The efficiency of the system is investigated through the simulations

View full abstract

We propose a novel classification method using a learning vector quantization, which utilizes not only the feature vectors but also the neighborhood information on data observation points. The proposed method achieves an effective classification even if the boundary of each class is overlapped in the feature space. The effectiveness of the proposed method is verified by applying it to the tissue classification problem of the intravascular ultrasound radiofrequency data.

View full abstract

The slow spread of knowledge management systems is caused by the large amount of data that must be acquired and learned before such systems can be used. Therefore, we have developed a new knowledge library inspired by two ideas: the Radio Frequency IDentification (RFID) tag system and simple 3-step rules. The former registers the processes of exchanging knowledge without troublesome manipulation. The latter encourages users to frequently exchange knowledge by following three steps: searching for a partner, exchanging knowledge mediated by books or papers, and recording this exchanging knowledge process. This paper reports our proposed knowledge library that supports new search criterion by using a human network as an index.

View full abstract

The paper introduces a nonparametric method of the regression function estimation based on a Takagi-Sugeno fuzzy model. The main feature of this method is the application of an information criterion as the sample objective function to be minimized, so that generalization capabilities of the estimator be enhanced. This optimization task is performed by a genetic algorithm preserving interpretability of the resulting model.

View full abstract

Harary's structural balance theory based on the idea of Heider explains our social process and is used to account for the social actors attitude toward another. We propose a new knowledge emergence system in the form of an agentbased simulation model. In this model, the structural balance theory is used to explain the feelings, attitudes and beliefs. Some communication patterns are also introduced for agents in the simulation. Each agent is in effort to reach balanced states and communicates with each other. We analyze a social value emergence by using the proposed model.

View full abstract

Visual object tracking is required by many vision applications such as human-computer interfaces, human-robot interactions and so on. However, in general living spaces where some of such applications are assumed to be used, stable tracking is generally difficult because there are many objects which are possible to cause the visual occlusion. Especially, conventional tracking techniques cannot deal with a complete occlusion over a long time instead of a short-time or partial occlusion. They also cannot handle the case that an occluder such as a box and a bag contains the tracking target and they move as one body. In this paper, to handle these problems, we propose a novel method for visual object tracking using a particle filter, which switches tracking targets autonomously. In our method, when an occlusion occurs in tracking, the model of the occluder is dynamically created and tracking target is switched to this model. Since the original target model is also stored simultaneously, it can return to the tracking target when the original target appears again from behind the occluder. We show the effectiveness of our method through some experiments.

View full abstract

A fuzzy constraint satisfaction problem (FCSP) is an extension of the classical CSP, a powerful tool for modeling various problems based on constraints among variables. Basically, the algorithms for solving CSPs are classified into two categories: the systematic search (complete methods based on search trees) and the local search (approximate methods based on iterative improvement). Both have merits and demerits. Recently, much attention has been paid to hybrid methods for integrating both merits to solve CSPs efficiently, but almost no attempt has been made so far for solving FCSPs. In this paper, we present a hybrid, approximate method for solving FCSPs. The method, called the Spread-Repair-Shrink (SRS) algorithm, combines a systematic search with the Spread-Repair (SR) algorithm, a local search method recently developed by the authors.

View full abstract

How to adapt agents to dynamic environments has been a crucial issue in the machine learning field. Many reinforcement learning (RL) methods have been proposed to address the above problem. The data obtained while robots acquire policies by RL are generally immediately revoked, or are eliminated shortly after a part of them is used to share the reward; however, they are considered to have potentials to reflect the characteristics of a correspondent environment. We have leveraged these data in the form of Bayesian Network (BN), and have proposed a system improving RL agents' policies with a mixture model of BNs. Our system allows agents to improve their policies by the information derived from the mixture model. This study aims to investigate the adaptability of our policy- improving system to dynamically-switched ""actual"" environments, and the relationships between the knowledge representations in the mixture and the policy in correspondent environments.

View full abstract

In this paper, we propose a frame rate up-conversion (FRUC) method using the motion compensation (MC) based on block matching algorithm (BMA). The property of the BMA is changed by the size of block. It is desirable for small moving object to set block size small. On the other hand, we would like to set the block size large in the global motion region. Therefore, the block size is decided depend on the property of the local region of frame image. In this paper, we present a novel FRUC technique based on the motion compensated interpolation with multi-size blocks. At first, the frame image is divided into 8x8 blocks and motion vectors are estimated for each 8x8 block. Next, the similar motion blocks are merged into one block. Thus the proposed method reduces block artifacts and realizes the clear interpolation of moving objects regardless of the size of those.

View full abstract

Applications as visual navigation with image sensor and mixed/augmented reality have been investigated actively. In these technique requires the self-location information of the camera. In this paper, we propose a method for the estimating camera position from outdoor image sequences by using the simple GPS information as first camera position. In this method, the camera position is estimated by responding corners of the building on the map to those on the image. And, the error on the image is prevented by matching borders between ground and building on the map and those on the image. In the experimental results, the evaluation of estimation of the camera position and the comparison the input image and the image which correspond to the estimated camera parameter show the efficacy of the proposed method.

View full abstract

A noise reduction method is proposed to reduce background noise in noisy speech. We have investigated a noise reconstruction system (NRS) based on a linear prediction error filter (LPEF) and a noise reconstruction filter (NRF). An input signal of a LPEF becomes a white signal. Assuming that background noise is generated by exciting a linear system by a white signal, a NRF can reconstruct the background noise from white noise by estimating a noise generating system. However, since a residual speech is included in an input signal of a NRF, it is necessary to use a small step size, which enables us to update coefficients of a NRF without estimating a speech signal. Thus, it is difficult for a NRS to track non-stationary noise. In order to solve the problem, a noise reconstruction system using step size control is proposed. The step size control takes advantage of cross-correlations between input signals and an enhanced speech signal.

View full abstract

When a speech recognition system is used in real environment, noises have a serious problem in the system. In this report, we propose an adaptive method modified from Running Spectrum Filtering/Dynamic Range Adjustment (RSF/DRA) techniques which normalize feature parameters after filtering the time trajectories of both the power spectra and the log power spectra. In order to improve the recognition accuracy, the designed method requires considerably high order FIR filter. Using a FRM filter, we can develop the robust speech recognition whose recognition ability is high enough and whose filtering order can be reduced low. As a result of the isolated word recognition experiment where fifteen kinds of noises were added, higher recognition rates was obtained than conventional method, i.e, RSF/DRA, Spectrum Subtraction(SS) and RelAtive SpecTrAl(RASTA).

View full abstract

Clustering has been used as a popular technique for finding groups of genes that show similar expression patterns under multiple experimental conditions. Many clustering methods have been proposed for clustering gene expression data, including the hierarchical clustering, k-means clustering, and selforganizing map. However, the conventional methods are limited to identify different shapes of clusters because they use a fixed distance norm when calculating the distance between genes. The fixed distance norm imposes a fixed geometrical shape on the clusters regardless of the actual data distribution. Thus, different distance norms are required for handling the different shapes of clusters.

View full abstract

In this paper, we elaborate MsS Sensor technology to the following point of views: Background of MsS, Sensor using Guided wave technology and features, MsS Sensor theory and guided wave. And we explain how to analyze signals reflected from the defects in the pipe structure using wavelet. In this paper, we validate that MsS sensor is useful to inspect and monitor a large area of the structure from a fixed test location.

View full abstract

To efficiently support mobility is one of the most challenging problems towards seamless integration of future wireless networks. It becomes increasingly important to support IP mobility as surprisingly explosive growth of Internet and impressive advances of the mobile communication technology. Mobile IP is the most widely known mobility management standard in IP based wireless networks. However, some inherent drawbacks of this protocol are partially solved. Although Route Optimization tries to solve these problems, it cannot manage mobility satisfactorily efficiently. In this paper, an intelligent mobility binding cache management scheme namely IBC (Intelligent Binding Caching) is proposed. The simulation results show that IBC proposed in this paper significantly enhances the overall performance of mobility management in IP based wireless networks.

View full abstract

We introduce the novel concept of ""Secure Spaces"", physical environments in which any resource is always protected from its unauthorized users' eyes or ears by assuredly enforcing its access control policies for pairs of it and each user inside them. Aiming to build such secure spaces, this paper proposes a model and an architecture for space entry control based on its dynamically changing contents, such as users, physical resources and virtual resources outputted by embedded devices. We firstly formalize the content-based entry control model and mechanism, and then describe the architecture for building secure spaces.

View full abstract

This paper presents a novel information processing technique called scale shrinking transformation (SST). SST comprises three steps: initialization, matrix transformation, and using the column vectors of the transformed matrix as the new input vectors. The essence of SST is that the structural correlation between original inputs can be obtained. More significantly, the transformed matrix contains elements with much smaller scale variation. When applied to existing feedforward neural networks, it can alleviate problems commonly encountered in tasks of function approximation, separating nonlinearly classes, and noise filtering. When the column vectors are used as the new input to a feedforward network that comprises hidden layers, training speed can be reduced. The input scale divergence problem that plagues higher-order neural networks can also be alleviated with SST.

View full abstract

In the real world application, recognition of two-dimensional images regardless of their rotational orientation and size, i.e., smaller or larger, is one of the significant problems in computer vision. Techniques of third order neural network and Zernike moment have shown to be successful in solving the problem, but their limitaion is costly in term of computational time and network complexity. In this paper, we apply a technique of Fuzzy c-Mena to resolve the problem of invariant recognition under rotation and scaling regardless of its vertical and horizontal size of an image. The learning of Fuzzy c-Mean does not have the invariant capability; therefor we presented in this paper a new technique with some modificaiotns based on the concept of principal component analysis.

View full abstract

Predicting failure of financial institutions can have a significant impact on the economy. Early warning systems developed from failure prediction models have proven to reduce as much business bankruptcy as possible. The recent East Asian economic crisis is a great lesson one should learn from lacking effective early warning systems. To serve as a sound early warning signal, the accuracy of a failure prediction model is as important as its robustness over time to failure. For an emerging market economy where ownership concentration is common, we show that only financial variables are not sufficient to produce models with a good predictive power. In our logit models that also incorporate ownership variables, 85.45%, 85.41%, and 91.49% of financial institutions are correctly classified in the models that used the data of one, two, and three years prior to the failure, respectively. In neural network model, the classification accuracy of a testing set is equal to 90.91% for one-year-ahead bankruptcy forcasting.

View full abstract

An ontology-based approach to context modeling is proposed to assist the development of context-aware applications in smart living space. In the proposed approach, the ontology-based context model is employed to collect, aggregate, store, share and reason the useful context information. Besides, to overcome the problems caused by the uncertain context information, such as inaccurate reasoning, a context quality model is also proposed to evaluate the credible degree of the collected context information. In the context quality model, a context quality metric is introduced to solve this problem and ensure the quality assurance of context information. The feasibility and effectiveness of the proposed models are verified through the implementation of a context-aware application in smart living space.

View full abstract

In this paper, we propose a technique of vector conversion of word and linguistic information processing neural network (LIPNN) with abilities of memory, recall, and inference of sentences expressed by semantic network form as one of the methods to treat natural language. The LIPNN receives Japanese texts, and outputs semantic network form of triple representation. First, the proposed network extracts a relation between words by CaboCha (Japanese dependency structure analyzer) and converts a natural language into a semantic network form of triple representation. Then thesaurus is used to convert a word into vector expression and input it into the LIPNN. Because of a word converted into vector expression, the proposed network is able to handle the word that does not have learned. The LIPNN is a three-layered structure to mem-ory and recalls triple representations and sentences. And also it can infer without control from the outside by newly proposed inhibit activation method. In the computer simula-tion, we carried out two kinds of simulations to confirm the network is able to handle the unlearned words and abilities of memory, recall, and inference of sentences in the network.

View full abstract

This paper introduces a new method which enables us to calculate the Sound Intensity Level (SIL) in dB near the pianist head in any moment of the MIDI file stream for the solo piano instrument. This method considers the characteristics of the piano sound and suggests a relatively simple algorithm to find SIL. Points to enhance are shown below and a further study is discussed as well.

View full abstract

This paper discusses the temporal coding in a gesture recognition method used for communication of a robot with a human. The proposed method is composed of a steady-state genetic algorithm for human hand detection, a spiking neural network for spatio-temporal pattern extraction of a human hand motion, and a self-organizing map for gesture clustering. In this paper, we discuss the effectiveness of the temporal coding of the spiking neural network through several experimental results.

View full abstract

A survey is done of various results in the control of bilateral teleoperation systems in order to determine the best control scheme for telesurgery over the internet. Telesurgery is a particularly demanding application where the challenges imposed by the internet, including time-varying delay, lost packets and out-of-order packets, must be addressed to try to maximize performance. The nature of delay on IP networks and the merits of UDP as a transport protocol are investigated and out-oforder packets are shown to have the same effect as lost packets. Stochastic models can be used to characterize packet delay. Passivity-based, predictive, robust & optimal and event-based control are discussed with respect to teleoperation. When timevarying delays are introduced, the system becomes a hybrid event-time based system. Future work is proposed that would consider a delay model that incorporates knowledge of lost and out-of-order packets to be incorporated into a hybrid event-time predictive control scheme.

View full abstract

A new accurate study method for vibration and noise analysis of gear structure body is presented based on the theory of elastomeric dynamics according to the theory of thick-walled plate and considering its actual working conditions and the influence of the direction of gear thickness. It takes an advantage of accurate analysis the gear body and has some merit to dynamic design of gear system, and the vibration displacements of the elastomer at three directions are got out. The accurate analytical expression of both natural frequency and natural functions (mode shapes) of the gear elastomer are given out. The three-dimensional shape and the three-dimensional distribution stress of the vibration model are presented in every point of gear elastomer for different parameter using FEM . It presents 8.47% precision improving compare with the analysis results of thin-walled plate theory. The theoretical analysis results are contrasted with numerical simulation results and the error shows as 6.16%. The proposed method brings an increase of about 18% in average acoustic radiation estimate precision and can be used to studying the characteristics of acoustic radiation and design of reducing noise of gears.

View full abstract

The concept of grey system theory, which has a certain prediction capability, offers an alternative approach for various kinds of conventional control methods, such as PID control and fuzzy control. For instance, grey PID type fuzzy controller designed in this paper, can predict the future output values of the system accurately. However, the forecasting step-size of the grey controller determines the forecasting value. When the step-size of the grey controller is large, it will cause over compensation, resulting in a slow system response. Conversely, a smaller step-size will make the system respond faster but cause larger overshoots. In order to obtain a better controller performance, another fuzzy controller is designed for changing the step-size of the grey controller. The value of the forecasting step-size is optimized according to the values of error and the derivative of the error. Moreover, the output of the grey controller is uptaded using the prediction error for better controller performance. It is clear that the proposed adaptive PID type fuzzy controller is effective in controlling such a non-linear system by changing the prediction horizon simultaneously for real-time working.

View full abstract

This paper presents a direct torque control (DTC) of a permanent magnet synchronous motor (PMSM), using DTC technique and space vector modulation (SVM). In the conventional DTC a hystersis controllers are used to limit the flux and torque in the control band while in the SVM DTC the controllers are replaced by PI or PI FLC controllers. The comparison shows that the torque dynamic response is very rapid and the system achieves the steady status in a very short time. The performance for two different motor rating low and medium power confirm that the proposed control strategy is good.

View full abstract

Social simulation is becoming more important tools in our society and organizations. An agent based simulation make it possible to construct bottom-up models for social and organizational systems. A social simulation dose not gives a universal natural law. However, it helps us to construct a shared social reality or internal model that is more realistic and complex but more intuitive and understandable compared with simple theoretical social models. The shared internal model as a social simulation model helps us collaborate by coordinating our decision making process. For social and organizational planning and problem solving, a social simulation model will become more important in the context of social and organizational architecture design in near future. We have to construct suitable social or organizational internal models to project our social and organizational activities under the open social and organizational architecture design principles. An agent based social simulation should be designed as a functional model that consists of many bottom-up social and organizational component modules. The modules should not be designed as black boxes. It should be designed clear bottom-up component. More over social simulation should help our understanding for the system. For the purpose, we have developed new social simulation language and modeling methodology. We have developed new social simulation language called SOARS. SOARS originally represents Spot Oriented Agent Role Simulator. Now It also represents Social and Organizational ARchitecture Simulator. SOARS consists of three layers of simulation environments. The top layer consists of the visual shell that help to build icon based intuitive modeling of social simulation. The visual shell also helps component-based programming for social simulation. The middle layer supports the SOARS simulation engine called the model builder. The model builder uses SOARS scripting language for describing simulation process. An icon based programming by the visual shell is translated to SOARS scripting language by the visual shell and executed by the model builder. New types of modular programming style by using new concepts such as stage and role are used in SOARS scripting language. The bottom layer provides tools for helping understanding and communicating processes for the social simulation. Animator helps animating and analyzing a simulated result. The gaming environment helps construct hybrid simulation that supports communicating and understanding processes. Where human players can attend the simulation with machine agents. We also support grid computing for SOARS that is indispensable for landscape search and model mining. We are developing new social simulation schema for social architecture design with the development of SOARS.

View full abstract

Many studies have been actively carried out in a distributed processing environment by using agent systems for efficient network management. Multi-agent systems promote the efficiency in solving problems by cooperating among agents. Also, each agent independently manages its own tasks by dividing a whole work into smaller units and assigning them to each agent. There are many application areas in the real world using the multi agent systems. One of these areas is a digital library system. The introduced literatures about the personalized digital library systems consider the topic of papers for analyzing the users' preferred papers. In this case, they can not recommend the papers whose topic is not relevant to the queried keywords request by a user but contents are relevant to them. In order to solve the problem, I consider the abstracts in the papers for providing a personalized paper search list according to the users' behavior on the papers and the relevance among keywords in the abstracts. Also, the problem from the multi agent-based digital library systems is that users themselves should visit all possible search servers one by one. To overcome the problems, a new platform of multi-agent digital library system which is mobile search system is proposed here. Users do not need to visit all possible search servers with the same query by using the new platform. The proposed system automatically visits to all possible search servers when user requests a query. In this case, the scheduling of visiting the search servers is needed. Also, the negotiation for the results searched from the servers needs to be made for complicated situations such as the duplicated search results from multiple servers. Also, the proposed personalized paper search algorithm builds users' individual relevance network from analyzing the appearance frequencies of keywords in the searched papers. The relevance network is personalized by providing weights to the appearance frequencies of keywords according to users' behaviors on the searched list, such as 'downloading,' 'opening,' and 'no-action'. From the experimental results, it is demonstrated the proposed method using 100 faculties' search information employed in the University of Suwon. Also, the performance of the proposed method is compared with that of the conventional paper search system by surveying the satisfaction of users for both systems.

View full abstract

Induction motors play a very important role in the safe and efficient operation of industrial plants and processes. The unexpected fault of motor causes many troubles such as the pause of overall process machinery as well as motors. In recent years, the fault detection and diagnosis of induction motors have been gaining more interests in the filed of highly reliable systems. For reliable fault diagnosis, it is extremely required in detecting and classifying the fault elements. There are some detection methods to identify the motor faults. Among the detection methods, the mainly used approaches are vibration monitoring and motor current signature analysis (MCSA). The vibration method is based on detecting vibration signal when motors happen to fault. This method, however, has some problems such as selection of reliable sensors and position attached on induction motors. For the fault diagnosis of three-phase induction motors, we set up an experimental unit and then develop a diagnosis algorithm based on pattern recognition. The experimental unit consists of induction motor drive and data acquisition module to obtain the fault signals. As the first step for diagnosis procedure, preprocessing is performed to make the acquired current simplified and normalized. To simplify the input data, three-phase currents are transformed into the magnitude of Concordia vector. As the next step, feature extraction is performed by PCA. Finally, we used the SVM classifier for fault detection of induction motor. To show the effectiveness, the proposed fault diagnostic system has been intensively tested with the various data acquired under the different electrical and mechanical faults with varying load.

View full abstract

In this paper, several applications based on Speech Signal Processing are introduced. The first study aims at developing a dialog system to improve the quality of life of the elderly with a dementia. The evaluation results showed that the dialog system was more responsive in catering to the needs of dementia patient than professional caregivers. The second study presents the new approach of identifying human emotional states such as anger, happiness, normal, sadness, or surprise in speech signals. The simulation performance showed that the recognition rates of vocal emotion gradually increased with an increase of adaptation sample number. The final application is based on the user-customized interaction using speech recognition, video processing, and fingerprint verification, for intelligent home environments. In evaluation, the results showed that the proposed system was easy to use for intelligent home environments, even though the performance of the speech recognizer was not better than the simulation results owing to the noisy environments.

View full abstract

A multi-resolution image similarity model based on region-based image similarity modeling and fuzzy aggregation operators is presented, where the overall image similarity between two images is based on fuzzyin three sets of crisp valued similarities: feature, region and image, respectively, in a hierarchical manner. It helps reducing the influence of inaccurate image segmentations of the global and region-based image similarity models. Compared with the image similarity modeling on either global or region-based representation with crisp valued feature, region or image similarity representations, the proposed modeling results in the better overall retrieval performance with an average retrieval precision higher between 2% and 6%. Compared to two image retrieval systems, SIMPLicity and WBIIS, the proposed model brings an increase of 2% and 22% respectively in average retrieval precision. The descriptive power of the image similarity model increases by allowing model itself to capture the variety of the similarity criteria when compared to the conventional image similarity models. The proposed multiresolution image similarity modeling is thus more suited when approximating human perception of the image similarity in image retrieval.

View full abstract

Dyadic Curvelet transform (DClet) is proposed as a tool for image processing and computer vision. It as an extended Curvelet transform solves the Curvelet inconvenience of decomposition into components at different scales. It provides simplicity, dyadic scales, and absence of redundancy for analysis and synthesis objects with discontinuities along curves, i.e., edges via directional basis functions. The performance of the proposed method is evaluated by removing Gaussian, Speckles, and Random noises from different noisy standard images. Average 24.79 dB Peak Signal to Noise Ratio (PSNR) compared to 23.63 dB via the wavelet is evidence that the DClet outperforms the wavelet for removing noise. The proposed method is robust that makes it suitable for biomedical world. It is a candidate for gray and color image enhancement and applicable for compression/ efficient coding which critical sampling might be relevant.

View full abstract

2D gel image maybe scanned at different resolutions or sizes for a single sample which would require readjustments so that the biologist can make more accurate analysis. However, the readjusting process could cause loss to some of the protein spots. The proposed scheme is to use the multiple resolution images on two methods. One method takes the minimal image as reference image and uses it to get the difference values of all the pixel values of its adjacent image. Bit map tables and difference value tables are created for all the adjacent images which require less storage and compressed a second time to free up more storage. The second method used is to apply the JPEG2000 compression to images. An image with resolution in between any two images in the list of multiple resolution images can be recovered with imperceptible distortions. Results show that more protein spots can be retained.

View full abstract

In this paper we first give a brief discussion of the newly emerging JPSEC security standard for JPEG 2000 compressed image. We then propose a scalable access control scheme specially designed for JPEG 2000 encoded images. Our scheme applies SNOW 2 stream cipher to JPEG 2000 codestreams in a way that it is possible to truncate the JPEG 2000 codestream after encryption, during transmission, storage, or by a third party unaware of the encryption key, and still be possible to decrypt the image to view it at a lower resolution or a quality level. We propose three implementations for our algorithm. Packet Based Access Control in which individual packet bodies are encrypted using the same encryption key and provides near optimum Rate Distortion truncation. Resolution Based Access Control and Quality Based Access Control in which each entire resolution or quality layer is encrypted using a different encryption key and provide controlled access to various resolutions or quality layers depending on the level of the key given to the user. Encryption keys are dependant of each other such that the key for a lower resolution or quality layer is the result of hashing the key of the higher resolution or quality layer. The hashing function used is the SHA-256. Our access control scheme preserves the inherent flexibility and accessibility of JPEG 2000 encoded images, and also preserves end-to-end security of encrypted images.

View full abstract

In this paper an optimal fuzzy controlled is proposed for the automation of a bus suspension system. First it is implemented a fuzzy controller to suspension system to observe its response, then it is analyzed to determine which parameters have influence in its behavior to be optimized with a simple algorithm. This will improve the suspension system and will give a better response for different road condition providing a good comfort to the rider.

View full abstract

A methodology for realizing energy and space saving adaptive control algorithms is proposed. The principle is based on the methodology of Pulse Width Modulation (PWM) with modified delta form. The dynamic range of the calculation becomes twice or more the conventional method. This methodology is very useful for realizing the embedded real-time intelligent systems or micro-electromechanical systems(MEMS), since complicated control algorithms can be implemented in fixed-point arithmetic with short word length. Finally, the simulated results are presented to show the practicality and effectiveness of the proposed implementation methodology.

View full abstract

This paper suggests novel hybrid optimization system (BF-CL) based on the bacterial foraging and clonal selection of immune system. A foraging strategy involves finding a patch of optimal condition (e.g., group of objective with conditions), deciding whether to enter it and search for optimal conditions, and when to leave the patch. There are predators and risks, energy required for optimization, and physiological constraints (sensing, memory, cognitive capabilities). Foraging scenarios can be modeled and optimal policies can be found using, for instance, dynamic programming. This approach provides us with novel hybrid model based on foraging behavior and clonal selection for a higher running time and with also a possible new connection between evolutionary forces in social foraging and distributed nongradient optimi-zation algorithm design for global optimization over noisy surfaces for control system.

View full abstract

Minimizing power consumption of inverter driven pumps is highly significant for drinking water treatment plants. This paper proposes a systematic optimization method based on topological and geometric properties of the objective function for obtaining multiple local optimal solutions of the inverter driven pump system. For the purpose of illustrating the proposed method, the pump system is studied numerically. Multiple local optimal solutions are obtained, and one of these solutions leads the lowest active electric power consumption. The numerical result shows the effectiveness of the proposed method with practical data of pumps.

View full abstract

Studies on the effectiveness of fuzzy logic for nonlinear modeling is presented. Although successful applications of fuzzy logic in many diverse areas are reported in the literature, commonly the explicit description of the key features of fuzzy logic yielding these outcomes is not addressed. The present research addresses this issue, to understand the basic features of fuzzy logic in nonlinear modeling applications in the context of dynamic system modeling, data modeling, signal modeling etc. This is accomplished by stochastic inputs to a nonlinear system modeled by fuzzy logic. The modeling performance is investigated in relation to the degree of nonlinearity of the model and the probability density of the stochastic inputs.

View full abstract

We suggest a method to detect rear car plate of one more cars by using blobs. First, we try to search all of the blobs from an input image based on the difference between objects and background. Second, we obtain rectangles enclosed the blobs, and rectangle clusters by considering the properties, for example, the number, size, distance, position. Third, the cluster is verified by the Support Vector Machine. Even if we only use the adaptive binarization as the preprocessing, the detection ratio is very high.

View full abstract

TAM(Topographic Attentive Mapping) network is a biologically-motivated neural network. It is composed of input layer, category layer, and output layer. The TAM network with three pruning rules for reducing links and nodes at the layer is called fuzzy TAM network. In this paper, we apply fuzzy TAM network to pattern analysis on core competency model for subcontractors of construction companies and show its usefulness.

View full abstract

PD(Partial discharges) are small electrical sparks that occur within the electric insulation of cables, transformers and windings on motors. PD analysis is a proactive diagnostic approach that uses PD measurements to evaluate the integrity of this equipment. Recently, several diagnostic algorithms for classifying the type of PD and locating the defect position have been developed. In this work, a new PD recognition system is proposed, which utilizes approximate coefficients of wavelet transform as a feature vector, furthermore, introduces SOM(Self-Organizing Map) to classify the various PD phenomena.

View full abstract