Search

In this paper the fusion of artificial neural networks, granular computing and learning automata theory is proposed and we present as a final result ANLAGIS, an adaptive neuron-like network based on learning automata and granular inference systems. ANLAGIS can be applied to both pattern recognition and learning control problems. Another interesting contribution of this paper is the distinction between pre-synaptic and post-synaptic learning in artificial neural networks. To illustrate the capabilities of ANLAGIS some experiments on knowledge discovery in data mining and machine learning are presented. Previous work of Jang et al. [1] on adaptive network-based fuzzy inference systems, or simply ANFIS, can be considered a precursor of ANLAGIS. The main, novel contribution of ANLAGIS is the incorporation of Learning Automata Theory within its structure. [ABSTRACT FROM AUTHOR]

The integrated technology of the artificial neural network is a research focus of the neural computing technology, which possesses ripe applications in a lot of fields. The neural network ensemble studies the same question with limited neural networks. The output of the ensemble under some input example is determined by all the output of the neural network forming the ensemble under the same input example. The negative correlation learning, which encourages different individual network to study and train different parts of the ensemble in order to make the whole ensemble study the whole training data better, is a training method for the neural network ensemble in this paper. Using a BP algorithm with impulse in the error function is an improvement of the method of negative correlation learning in the paper. The method is an algorithm in batches with more powerful generalization ability and studying of speed, because it combines primitive correlation learning with BP algorithm of impulse. [ABSTRACT FROM AUTHOR]

This paper presents a comparison of the effectiveness of two computational intelligence approaches applied to the task of retrieving rhythmic structure from musical files. The method proposed by the authors of this paper generates rhythmic levels first, and then uses these levels to compose rhythmic hypotheses. Three phases: creating periods, creating simplified hypotheses and creating full hypotheses are examined within this study. All experiments are conducted on a database of national anthems. Decision systems such as Artificial Neural Networks and Rough Sets are employed to search the metric structure of musical files. This was based on examining physical attributes of sound that are important in determining the placement of a particular sound in the accented location of a musical piece. The results of the experiments show that both decision systems award note duration as the most significant parameter in automatic searching for metric structure of rhythm from musical files. Also, a brief description of the application realizing automatic rhythm accompaniment is presented. [ABSTRACT FROM AUTHOR]

This paper proposes a neural network-based classification approach to inventory risk level of spare parts. Firstly a fuzzy evaluation of spare parts is made in terms of their availability of suppliers, importance, predictability of failure, specificity and lead time. Then a multilayer feed forward neural network model is established. The Back Propagation (BP) algorithm for training a neural network is used to decide the weights to connections in the model. Choosing a sample of historical data of 100 spare parts and undertaking a BP training stimulation, the model is used to predict the inventory risk levels of 60 spare parts for a well-logging service firm. The forecasting reliability reaches 84%. [ABSTRACT FROM AUTHOR]

This paper depicts and evaluates an evolutionary design process for generating a complex self-organizing multicellular system based on Cellular Automata (CA). We extend the model of CA with a neural network that controls the cell behavior according to its internal state. The model is used to evolve an Artificial Neural Network controlling the cell behavior in a way a previously defined reference pattern emerges by interaction of the cells. Generating simple regular structures such as flags can be learned relatively easy, but for complex patterns such as for example paintings or photographs the output is only a rough approximation of the overall mean color scheme. The application of a genotypical template for all cells in the automaton greatly reduces the search space for the evolutionary algorithm, which makes the presented morphogenetic approach a promising and innovative method for overcoming the complexity limits of evolutionary design approaches. [ABSTRACT FROM AUTHOR]

Over past decades, Artificial Neural Network (ANN) area has been the focal point of an ever-increasing number of research works and a very active pivot of interdisciplinary research activity. It is now time to state if ANN are ready to defeat nowadays΄ real-world and industrial challenges. The main goal of this paper is to present, through some of main ANN models and based techniques, their capability in real world industrial dilemmas solution. Examples of real world and industrial applications have been presented and discussed. [ABSTRACT FROM AUTHOR]

Over the recent years, new public security tendency to fit up public areas with biometric devices has emerged new requirements in biometric recognition dealing with what we call here ˵mass biometry″. If the goal in ˵individual biometry″ is to authenticate and/or identify an individual within a set of favored folks, the aim in ˵mass biometry″ is to classify a suspect individual or behavior within a flow of mass customary information. In this case, the ability of handling relatively poor information and the skill of high speed processing become chief requirements. These antagonistic requests make the ˵mass biometry″ and related applications among the most challenging frames. In this paper we present an ANN based system in a ˵mass biometry″ context using facial biometric features. The proposed system takes advantage from kernel functions ANN model and IBM ZISC based hardware. Experimental results validating our system are presented and discussed. [ABSTRACT FROM AUTHOR]

This paper covers a comparison between two distinct approaches to neural network modeling. The first one is based on a developing of a single neural network model to predict bad debt events. The second one is based on combined models, building firstly a clustering model to recognize the pattern assigned to the customers, with a particular focus on the insolvency, and then developing several distinct neural networks to predict bad debt. In the second approach, for each group identified by the clustering model one neural network had been constructed. In that way, we turned the quite heterogeneous customer base more homogeneous, increasing the average accuracy for the predictive modeling once several straightforward models were built. [ABSTRACT FROM AUTHOR]

In this paper we use a neural network approach for defects identification in textile. The images analyzed came from an artificial vision system that we used to acquire and memorize them in bitmap file format. The vision system is made of two grey scale line scan camera arrays and each array is composed of four CCD cameras with a sensor of 2048 pixels. Every single camera has a field of view of 600mm. The big amount of pixels to be studied to determine whether the texture is defective or not, requires the implementation of some encoding technique to reduce the number of the significant elements. The artificial neural networks (ANN) are manipulated to compress a bitmap that may contain several defects in order to represent it with a number of coefficients that is smaller than the total number of pixel but still enough to identify all kinds of defects classified. An error back propagation algorithm is also used to train the neural network. The proposed technique includes, also, steps to break down large images into smaller windows or array and eliminate redundant information. [ABSTRACT FROM AUTHOR]

Some features of a large number of combinatorial optimization problems prevent the use of exact solution methods, thus requiring the application of heuristic techniques to find good solutions, not always the optimal ones, in a feasible amount of time. This paper describes a heuristic approach, which is a hybrid between artificial neural networks and artificial immune systems, to solve the capacitated vehicle routing problem. This algorithm is based on a competitive model, which does not use a cost or evaluation function to determine the quality of the solution proposed. Despite this apparent drawback, the set of tests conducted with the proposed approach indicates a good performance of the algorithm when compared with similar works from the literature and the known best solutions available. [ABSTRACT FROM AUTHOR]

This paper deals with the topic of learning through neuroevolutionary algorithms in non-stationary settings. This kind of algorithms that evolve the parameters and/or the topology of a population of Artificial Neural Networks have provided successful results in optimization problems in stationary settings. Their application to non-stationary problems, that is, problems that involve changes in the objective function, still requires more research. In this paper we address the problem through the integration of implicit, internal or genotypic, memory structures and external explicit memories in an algorithm called Promoter Based Genetic Algorithm with External Memory (PBGA-EM). The capabilities introduced in a simple genetic algorithm by these two elements are shown on different tests where the objective function of a problem is changed in an unpredictable manner. [ABSTRACT FROM AUTHOR]

The aim of the research study presented in this paper is the automatic recognition of a singing voice. For this purpose, a database containing sample recordings of trained and untrained singers was constructed. Based on these recordings, certain voice parameters were extracted. Two recognition categories were defined – one reflecting the skills of a singer (quality), and the other reflecting the type of the singing voice (type). The paper also presents the parameters designed especially for the analysis of a singing voice and gives their physical interpretation. Decision systems based on artificial neutral networks and rough sets are used for automatic voice quality/ type classification. Results obtained from both decision systems are then compared and conclusions are derived. [ABSTRACT FROM AUTHOR]

In this paper, we present a framework used for creating, training, and testing SOM neural networks, which are used to recognize student learning styles under different pedagogical models. The SOMs are part of the student model of Intelligent Tutoring Systems we implemented for mobile devices and Web-based Learning Systems. The main contribution of this paper is the framework to build SOMs which can be used with any pedagogical model of learning styles. The SOM network produced with our framework has been tested with mobile devices and a system of web-based learning. [ABSTRACT FROM AUTHOR]

The assimilation of formal logic into the domain of Software Engineering offers the possibility of enormous benefits in terms of software reliability and verifiability. To date, however, the integration of such techniques has proved difficult since they involve a significantly increased burden on the programmer in meeting the demands of the formal mechanisms being employed. The current paper investigates the advantages which may be gained by the software development process with the introduction of Artificial Neural Network technology into a formal software development system. Essentially, the adaptive artificial neural network model is employed to refine an existing formal software model in order to produce increasingly better approximations to a given solution. Each approximation is itself a valid formal system whose precise behaviour may be formally determined. The paper introduces a framework by which a programmer may define a system possessing the abstract structure of a traditional neural network but whose internal structures are taken from the formal mathematical domain of Constructive Type Theory. The system will then refine itself to produce successive approximations to a desired goal based on data presented to it. An example is presented addressing a problem domain which has previously proved difficult to model. Although the example presented is necessarily limited, it does provide an insight into the potential advantages of merging formal logic with artificial neural systems. [ABSTRACT FROM AUTHOR]

- In this paper, the out coming temperature out of Claus reaction (which is used for recycling sulfur from H2S in refineries of Gas) is estimated via adaptive linear neuron networks. In order to get to the desired results, we need to be aware of the amount of H2S flow as well as the flow of the air. Acid gas used in the gas refinery involving in the prior reaction, has impurities such as: CO2, H2O, DMEA and stays null. Flow meters existing in the acid gas line movement, measure H2S and impurities (gas flow in complete). But for estimating the temperature of reaction and controlling it, we need the real amount of gas flow. In this paper, we have been trying to show that via using adaptive filter, we can separate impurity of acid gas from H2S. [ABSTRACT FROM AUTHOR]

This paper proposes a new technique using discrete wavelet transform (DWT) and artificial neural network for fault classifications on single circuit transmission lines. Simulations and the training process for the artificial neural network are performed using ATP/EMTP and MATLAB. The mother wavelet daubechies4 (db4) is employed to decompose, high frequency component from these signals. Positive sequence current signals are used in fault detection decision algorithm. The variations of first scale high frequency component that detect fault are used as an input for the training pattern. Back-propagation (BP) neural network is also compared with the RBF neural network in this paper. The result is shown that an average accuracy values obtained from RBF gives satisfactory results with less training time, and will be very useful in the development of a power system protection scheme. [ABSTRACT FROM AUTHOR]

Nowadays, we make use of the digital quantity to store and recover the information of the continuous quantity. It often takes us relatively large storage space to store the information, and can't acquire as much information of the continuous quantity as possible, because the traditional sampling process can't acquire the information of the adjacent sampling points' relationship. This paper puts forward a new idea that it take advantage of the structure of the Neural Network to store the information of the continuous quantity, so that it can greatly reduce the amount of storage space and recover as much information of the continuous quantity as possible. This paper gives an example of the application of Neural Network in The Technology of Image Processing. It tells us how to store and recover the image with the structure of Neural Network as a stored medium. The result show that, in this way, we can acquire a clearer image than using the traditional method, and greatly reduce the storage space which is used to store the information of the originally continuous quantity. [ABSTRACT FROM AUTHOR]

In recent years, Machine Learning (ML) has witnessed a great increase of storage capacity of computer systems and an enormous growth of available information to work with thanks to the WWW. This has raised an opportunity for new real life applications of ML methods and also new cutting-edge ML challenges like: tackle with massive databases, Distributed Learning and Privacy-preserving Classification. In this paper a new method capable of dealing with this three problems is presented. The method is based on Artificial Neural Networks with incremental learning and Genetic Algorithms. As supported by the experimental results, this method is able to fastly obtain an accurate model based on the information of distributed databases without exchanging any data during the training process, without degrading its classification accuracy when compared with other non-distributed classical ML methods. This makes the proposed method very efficient and adequate for Privacy-Preserving Learning applications. [ABSTRACT FROM AUTHOR]

This paper presents a multidisciplinary study on the application of statistical and neural models for analysing data on immissions of atmospheric pollution in urban areas. Data was collected from the network of pollution measurement stations in the Spanish Autonomous Region of Castile-Leon. Four pollution parameters and a pollution measurement station in the city of Burgos were used to carry out the study in 2007, during a period of just over six months. Pollution data are compared, their values are interrelated and relationships are established not only with the pollution variables, but also with different weeks of the year. The aim of this study is to classify the levels of atmospheric pollution in relation to the days of the week, trying to differentiate between working days and non-working days. [ABSTRACT FROM AUTHOR]

The maximum operating point of solar Photovoltaic (PV) panels changes with environmental conditions. Many methods have been proposed to locate and track the maximum power point (MPP) of PV cells. The difficulties that face these methods are the rapid changes in solar radiation and the variety in cell temperature which affects the MPP setting. External sensors are used in many approaches to measure solar irradiation and ambient temperature to estimate the MPP as a function of data measured. In this paper, linear correlation is proposed to analyze the experimental data to select the appropriate PV parameters that can recognize the MPP location. Short circuit current (Isc) and open circuit voltage (Voc) are selected as inputs factors instead of environmental influences. The paper demonstrates how these simple factors are necessary to locate accurate MPP under wide changes in environmental conditions. The statistical analysis is used to classify the data in appropriate fuzzy memberships. The proposed maximum power point tracking (MPPT) model depends on an Adaptive Neuro-Fuzzy Inference System (ANFIS) which is designed as a combination of the Sugeno fuzzy model and neural network. ANFIS of five layers with four fuzzy rules is used to acquire a high precision of locating Vmax with few adaptation epochs. The fuzzy logic controller (FLC) utilizes the ANFIS output voltage to track the MPP. The MPPT controller is designed to acquire high efficiency with low fluctuation. [ABSTRACT FROM AUTHOR]

Theory presented by Ashby states that the process of homeostasis is directly related to intelligence and to the ability of an individual in successfully adapting to dynamic environments or disruptions. This paper presents an artificial homeostatic system under evolutionary control, composed of an extended model of the GasNet artificial neural network framework, named NSGasNet, and an artificial endocrine system. Mimicking properties of the neuro-endocrine interaction, the system is shown to be able to properly coordinate the behaviour of a simulated agent that presents internal dynamics and is devoted to explore the scenario without endangering its essential organization. Moreover, sensorimotor disruptions are applied, impelling the system to adapt in order to maintain some variables within limits, ensuring the agent survival. It is envisaged that the proposed framework is a step towards the design of a generic model for coordinating more complex behaviours, and potentially coping with further severe disruptions. [ABSTRACT FROM AUTHOR]

An increasing effort has being devoted to researching on the field of Intrusion Detection Systems (IDS΄s). A wide variety of artificial intelligence techniques and paradigms have been applied to this challenging task in order to identify anomalous situations taking place within a computer network. Among these techniques is the neural network approach whose models (or most of them) have some difficulties in processing traffic data ˵on the fly″. The present work addresses this weakness, emphasizing the importance of an appropriate segmentation of raw traffic data for a successful network intrusion detection relying on unsupervised neural models. In this paper, the presented neural model is embedded in a hybrid artificial intelligence IDS which integrates the case based reasoning and multiagent paradigms. [ABSTRACT FROM AUTHOR]

This paper presents a method for co-evolving neuro-inspired developmental programs for playing checkers. Each player΄s program is represented by seven chromosomes encoding digital circuits, using a form of genetic programming, called Cartesian Genetic Programming (CGP). The neural network that occurs by running the genetic programs has a highly dynamic morphology in which neurons grow, and die, and neurite branches together with synaptic connections form and change in response to situations encountered on the checkers board. The results show that, after a number of generations, by playing each other the agents play much better than those from earlier generations. Such learning abilities are encoded at a genetic level rather than at the phenotype level of neural connections. [ABSTRACT FROM AUTHOR]

In this paper, a recently developed fiber tracking algorithm to be used with diffusion tensor (DT) fields acquired via magnetic resonance imaging (MRI) is improved and applied to real brain DT-MR images. The method performs satisfactorily in regions where branching and crossing fibers exist and offers the capability of reporting a probability value for the computed tracts. This certainty figure takes into account both the anisotropy and the information provided by all the eigenvectors and eigenvalues of the diffusion matrix at each voxel. In previous papers of the authors, a simpler algorithm was applied only to elementary synthetic DT-MR images. As now presented, this algorithm is now adequately used with more intricate synthetic images and is applied to real white matter DT-MR images with successful results. A novel neural network is used to estimate the main parameters of the algorithm. Numerical experiments show a performance gain over previous approaches, specially with respect to convergence and computational load. The tracking of white matter fibers in the human brain will improve the diagnosis and treatment of many neuronal diseases. [ABSTRACT FROM AUTHOR]

Although the current E-Learning systems have many merits, many of them only treat advanced information technology as simple communication tools, and release some learning contents and exercises in the network. In this paper, a one-class-in-one network for emotion recognition system in E-learning is proposed and implemented in the paper. Using a large database of phoneme-balanced Chinese words read by speakers consciously trying to portray an emotion, we trained and tested this module. We achieved a recognition rate of approximately 55%. The results obtained in this study demonstrate that emotion recognition in speech is feasible, and that neural networks are well suited for this task. [ABSTRACT FROM AUTHOR]

Recent years have witnessed intense research in the general area of Multi-Classifier systems (MCS), but this has rarely incorporated any utilisation of weightless neural systems(WNS) as the combiner of an MCS ensemble. This paper explores the application of weightless networks within the multi-classifier environment by introducing an intelligent multiclassifier system using a WNS called the Enhanced Probabilistic Convergent Neural Networks (EPCN). The paper explores the use of EPCN by illustrating its major features, such as the specification of disjoint or overlapping input subset to the MCS, and the inherently parallel nature of the design. Within the proposed system, the number of base classifiers per MCS could be specified manually or automatically. . The proposed MCS is problem-domain independent and, our investigation is performed on handwritten characters. The proposed MCS is adaptive, its combiner is capable of extracting absolute or weighted classification decision(output) from base classifier. Diversity is increased in the base classifier by injecting randomness into the system parameters. Two types of EPCN classifiers are proposed, fix-PCN and rand-PCN. These PCNs are independent and orthogonal. One uses a fixed method of forming connectivity while the other uses random method of forming connectivity. In order to verify the performance of the recognition system, tests were performed, off-line, on benchmark datasets of unconstrained handwritten numerals. Experimental results suggest that MCS outperforms single EPCN in classification of handwritten characters. [ABSTRACT FROM AUTHOR]

In this paper, a novel harmonic reduction technique in PWM AC voltage controller is proposed to reduce the harmonic contents in current waveform. Our technique is used to evaluate the optimal turn on and turn off angles in PWM such that the total current harmonic distortion is minimized. To accomplish this optimization tasks, in this paper, Genetic Algorithm (GA) is adopted. However, the evolved angles are only optimized at a given desired output voltage. To apply our proposed technique for all of output voltages, artificial neural network (ANN) is investigated to approximate the switching angles from sets of optimal angles evolved by GA. Simulation results show that our proposed technique is suited for designing and gains a better performance and reduction of current harmonics compared to conventional fixed-pulse PWM. [ABSTRACT FROM AUTHOR]

Wireless sensor node is composed of computational unit, sensing unit and a radio unit for communication, all embedded in a tiny unit. Maximum battery power is consumed by communication unit. Battery power is the prime source for wireless sensor node to function. Hence every aspects of wireless sensor network (WSN) are designed with energy constraints. ART1 and Fuzzy ART Neural Network models can be used very efficiently for developing Real time Classifier. Wireless sensor networks demand for the real time classification of sensor data. In this paper classification and clustering techniques using ART1 and Fuzzy ART is discussed. The proposed classifier can be a part of embedded microsensor. The paper discusses classification technique, which can reduce the energy need for communication. Three different clustering architecture are discussed which works at node level, as clustered group of nodes and for extracting several parameters with different sensitivity. [ABSTRACT FROM AUTHOR]

This paper discusses a Web-based parallel and distributed computing platform that supports multiple users to run experiments and optimizations with different deterministic/stochastic simulation systems. The platform is designed to be multi-tier client/server based in which all complex components, including various metaheuristic search algorithms, neural network based metamodels, deterministic/stochastic simulation systems and the corresponding database management system are integrated in a parallel and distributed platform and made available for general users to easily access, anytime, anywhere, through Web Services technology. In addition to revealing the system architecture of this platform, this paper also addresses the importance of logging simulation experiment results and optimization trajectories into the database to support advanced features like metamodelling, dynamically changing metaheuristic algorithms and fault tolerance within an integrated framework. Performance improvement on simulation optimization processes using these integrated features are illustrated with both multi-objective and single-objective optimization problems taken from industry. [ABSTRACT FROM AUTHOR]

The paper studied the cluster processing and efficiency on electrocardiogram waveform. 24 hours dynamic electrocardiogram waveforms were the analysis object. Analysis vectors were produced by wavelet transform, by which the feature of electrocardiogram was extracted. Four cluster methods were selected and compared in the paper on clustering efficiency and time complexity. According to the experiments result, on self-organization map (SOM) neural network, the basic waveform recognizing rate reached 77%, and accuracy rate reached 85%. It is the best in performance among others. On time complexity comparison SOM neural work will be limited in processing more object, it can be substituted by classification method that studied in other paper. [ABSTRACT FROM AUTHOR]

Recently, in digital distance relaying, different methods have been proposed for fault location estimation using one-end data and two-end data. The first method is more practical and economical since no communication channels and special devices are needed as in method "2". The first part of this paper proposes a new accurate approach for fault section estimation using probabilistic neural network based on the phase voltages and line currents phasor measurements from one end of the power transmission line. This approach takes the combined effect of fault resistance and load flow into consideration in the training data and assumes fault type is available. The second part of the paper suggests a recent method to estimate the error in voltage measurements at the relay point due to the existence of ground resistance and load current using generalized regression neural network (GRNN). Finally, modular SCGNN operates successively with PNN and GRNN to estimate the distance to the fault accurately at each section with the aid of positive, negative, zero sequence components of voltages and zero sequence current. The data required for the training of ANN is measured and processed using MATLAB Simulink. [ABSTRACT FROM AUTHOR]

Offset printing is a complex production craft. There are various factors which affect its quality. Decision control system of quality based on intelligent decision system which used case-based reasoning (CBR) and fuzzy neural network was proposed in the paper. It is a new way for intelligent decision of offset printing quality and control. This paper had described the systematic model and the key technologies such as case organization and case retrieval based on fuzzy neural network in detail. [ABSTRACT FROM AUTHOR]

This paper provides an overview of work undertaken over the past two years to develop Artificial Neural Network (ANN) techniques to improve the accuracy and reliability of road selection during map-matching (MM) computation. MM positions provided by low-cost GPS receivers have great potential when integrated with hand-held or in-vehicle Geographical Information System (GIS) applications, especially those used for tracking and navigation, on path and road networks. The applied modular neural network (MNN) approach is using a suitable road shape indicator to incorporate different road shapes for local ANN training. MNN test results indicate good potential for the method to provide a significant improvement in MM and positional accuracy over traditional methods. Further results and conclusions of this on-going research will be published in due course. [ABSTRACT FROM AUTHOR]

This paper presents a novel method that uses incremental self-organizing map (ISOM) network and wavelet transform together for the segmentation of magnetic resonance (MR), computer tomography (CT) and ultrasound (US) images. In order to show the validity of the proposed scheme, ISOM has been compared with Kohonen΄s SOM. Two-dimensional continuous wavelet transform (2D-CWT) is used to form the feature vectors of medical images. According to the selected two feature extraction methods, features are formed by the intensity of the pixel of interest or mean value of intensities at one neighborhood of the pixel at each sub-band. The first feature extraction method is used for MR and CT head images. The second method is used for US prostate image. [ABSTRACT FROM AUTHOR]

This paper presents a new neuro-based method for short term load forecasting of Iran national power system (INPS). A MultiLayer Perceptron (MLP) based Neural Network (NN) toolbox has been develeped to forecast 168 hours ahead. The proposed MLP has one hiden layer with 5 neurons. The effective inputs were selected through a peer investigation on historical data released from the INPS. To adjust the parameters of the MLP, the Levenberg-Marquardt Back Propagation (LMBP) training algorithm has been employed because of its remarkable fast speed of convergence. Most of papers dealt with 168-hour forecasting employed a hirachical method in the sense of monthly or seasonly provided that there are enough data. In the absence of rich data, forecasting error would increase. To remedy this problem, the proposed neuro-based approach uses only the weekly group data of concern while an extra input is added up to indicate the month. In other words for each weekly group, a unique MLP based neural network is designed for the purposed of load forecasting. [ABSTRACT FROM AUTHOR]

Artificial Neuron Network tool is adopted in an attempt to long term earthquake prediction for the Japan Trench subduction zone where an shock with magnitude 9 occurred last year and the probability model failed in forecast even after the shock. The preliminary result shows that the AI tool is helpful in such difficult a prediction, it can recognize some kind of rhythm of seismicity fluctuation that people can also find in the time series, but cannot be clear described. [ABSTRACT FROM AUTHOR]

Globalization and technological advancement has created a highly competitive market in the stock and share market industry. Performance of the industry depends heavily on the accuracy of the decisions made at performance level. The stock market is one of the most popular investing places because of its expected high profit. For prediction, technical analysis approach, that predicts stock prices based on historical prices and volume, basic concepts of trends, price patterns and oscillators, is commonly used by stock investors to aid investment decisions. In recent years, most of the researchers have been concentrating their research work on the future prediction of share market prices by using Statistical & Quantitative tools. But, in this paper we newly propose a methodology in which the Multiple Linear Regression and neural networks is applied to the investor΄s financial decision making to invest all type of shares irrespective of the high / low index value of the scripts, in a continuous time frame work. The proposed network has been tested with stock data obtained from the Asian Stock Market Database. Finally, the design, implementation and performance of the proposed multiple linear regression and model of simulated neural network are described. [ABSTRACT FROM AUTHOR]

The problem of using natural languages as a medium of input to computational system has long intrigued and attracted researchers. This problem becomes especially acute for systems that have to deal with massive amount of data as inputs in the form of sentences/commands/phrase as a large number of such phrases may look vastly different in lexical and grammatical structure but yet convey similar meanings. In this paper, we describe a novel approach involving Artificial Neural Network to sufficiently solve the aforesaid problem for inputs in English language. The proposed system uses Self Organizing Map (SOM) to recognize and classify the input sentences into classes representing phrases/sentences having similar meaning. After Detailed analysis and evaluation, we have been able to reach a maximum efficiency of approximately 92.5% for the system. The proposed expert system could be extended to be used in the development of efficient and robust systems like intelligent medical systems, Systems for Intelligent Web-Browsing, telemarketing and several others which will be able to take text input in the form commands/sentences in natural languages to give suitable output. [ABSTRACT FROM AUTHOR]

A field study was carried out in Yinchuan to gather and evaluate information about the real environment. O3 (Ozone), PM10 (particle 10 um in diameter and smaller) and SO2 (sulphur monoxide) constitute the major concern for air quality of Yinchuan. This paper addresses the problem of the predictions of such three pollutants by using the ANN. Because ANNs are non-linear mapping structure based on the function of the human brain. They have been shown to be universal and highly flexible function approximation for any date. These make powerful tools for models, especially when the underlying data relationship is unknown. [ABSTRACT FROM AUTHOR]

Breast cancer is the second leading cause of cancer deaths in women worldwide and occurs in nearly one out of eight women. Currently there are three techniques to diagnose breast cancer: mammography, FNA (Fine Needle Aspirate) and surgical biopsy. In this paper we develop an integrated expert system for diagnosis, prognosis and prediction for breast cancer using soft computing techniques. The basic aim is to compare the various neural network models from the recent literature. Breast cancer database used for this purpose is from the University of Wisconsin (UCI) Machine Learning Repository. Three different data sets have been used, each employing different diagnostic technique. It can use diagnosis, prognosis and survivability prediction of breast cancer patient in one intelligent system. We implement six models of neural networks namely Back Propagation Algorithm, Radial Basis Function Networks, Learning vector Quantization, Probabilistic Neural Networks, Recurrent Neural Network, and Competitive Neural network. Experimental Results show that different models give optimal performance for different types of data sets. However, all the models are able to solve the problem to a reasonable extent. [ABSTRACT FROM AUTHOR]

Artificial neural networks (ANNs) have been successfully applied to many areas due to its powerful ability both for classification and regression problems. For some difficult problems, ANN ensemble classifiers are considered, instead of a single ANN classifier. In the previous study, the authors presented the systematic trajectory search algorithm (STSA) to train the ANN. The STSA utilizes the orthogonal array (OA) to uniformly generate the initial population to globally explore the solution space, and then applies a novel trajectory search method to exploit the promising areas thoroughly. In this paper, an evolutionary constructing algorithm, called the ESTSA, of the ANN ensemble is proposed. Based on the STSA, the authors introduce a penalty term to the error function in order to guarantee the diversity of ensemble members. The performance of the proposed algorithm is evaluated by applying it to train a class of feedforward neural networks to solve the large n-bit parity problems. By comparing with the previous studies, the experimental results revealed that the neural network ensemble classifiers trained by the ESTSA have very good classification ability. [ABSTRACT FROM AUTHOR]