By Topic

Neural Information Processing, 2002. ICONIP '02. Proceedings of the 9th International Conference on

Date 18-22 Nov. 2002

Go

Filter Results

Displaying Results 1 - 25 of 115
  • Segmentation of pathological features in MRI brain datasets

    Page(s): 2673 - 2677 vol.5
    Save to Project icon | Click to expandQuick Abstract | PDF file iconPDF (670 KB) |  | HTML iconHTML  

    One of the major clinical applications of magnetic resonance imaging (MRI) is to detect pathological features in human body parts. While results are available in a digital format, their evaluation is performed by a trained human observer, which is still considered as the "gold standard". However, providing additional quantitative figures (e.g., lesion size or count) is tedious for a human and may better be obtained from automatical image processing methods. Three example brain lesion types (as revealed by MRI) and methods for their detection are described. Special emphasis is led on the way prior knowledge about the specific lesion type is incorporated in the algorithm. View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • Segmentation and recognition of on-line Pitman shorthand outlines using neural networks

    Page(s): 2454 - 2458 vol.5
    Save to Project icon | Click to expandQuick Abstract | PDF file iconPDF (513 KB) |  | HTML iconHTML  

    This paper presents a novel approach for the segmentation and recognition of the on-line vocalized outlines of Pitman shorthand. Due to its low redundancy, the recognition of the Pitman Shorthand requires high-performance outline segmentation and stroke classification. Our approach includes (1) the segmentation of the vocalized outlines, including the detection of over-segmentation using a neural network, (2) the recognition of Pitman shorthand consonant signs using another neural network, and (3) the word recognition based on the estimation of the overall confidence on the stroke classification. Experimental results on a small test set containing 68 most frequently used English words are reported in the paper. The average accuracy on these test words can reaches 89.6% by using our approach. View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • Application of PCA method to weather prediction task

    Page(s): 2359 - 2363 vol.5
    Save to Project icon | Click to expandQuick Abstract | PDF file iconPDF (592 KB) |  | HTML iconHTML  

    A method of short-term weather forecasting based on artificial neural networks is presented. Each training sample consists of a date information combined with meteorological data from the last three days gathered at the meteorological station in Miami, USA. The prediction goal is the next day's temperature. Prediction system is built based on multilayer perceptron network trained with backpropagation algorithm with momentum. The average prediction error of the network on the test set equals 1.12°C. The average percentage prediction error is equal to 5.72%. The results are very encouraging and provide a promise for further exploration of the issue. The so-called correlation ratio δ between predicted and real changes (trends) is equal to 0.7136. Relatively high value of δ additionally confirms good quality of presented results. Experimental results of application of the principal component analysis method at the stage of pre-processing of the input data are also presented. In that case the average prediction error and the average percentage prediction error are equal to 1.41°C and 7.93%, respectively. In order to explain the reasons of the poorer results obtained with the PCA method a closer look at the principal components defined by the network is presented. Possible reasons of the PCA failure are pointed out. View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • Design of neuro-fuzzy network for image compression

    Page(s): 2440 - 2443 vol.5
    Save to Project icon | Click to expandQuick Abstract | PDF file iconPDF (401 KB) |  | HTML iconHTML  

    The main objective of this paper is to propose a Neuro-Fuzzy based algorithm for Image compression. The inputs to the network are original image data, while the outputs are reconstructed image data, which are close to the inputs. If the amount of data required to store the hidden unit values and the connection weights to the output layer is less than the original data, compression is achieved. The compression ratio achieved in this paper is about 9 with good reconstructed image quality. The proposed network has an additional feature that each addition of a hidden unit to the network will always improve the image quality. Further the user can trade between image quality and compression ratio depending on the application requirement. The results are found to be better than the conventional methods. View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • Research on correct convergence of the EM algorithm for Gaussian mixtures

    Page(s): 2660 - 2664 vol.5
    Save to Project icon | Click to expandQuick Abstract | PDF file iconPDF (484 KB) |  | HTML iconHTML  

    In this paper, we present a theoretical analysis on the correct convergence of expectation-maximization algorithm for Gaussian mixtures. We first introduce the expectation-maximization algorithm and its general convergence properties. We also give a variation of the expectation-maximization algorithm for Gaussian mixtures. We then prove that the expectation-maximization algorithm becomes a compact mapping in certain neighborhood of a consistent solution when a measure of the average overlap of Gaussian in the mixtures is small enough while the sample size is large enough. We further obtain and prove the condition of the correct convergence of it. And finally, we demonstrate the theoretical results by the simulation. View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • Digital signatures and public key cryptosystems with multilayer perceptrons

    Page(s): 2308 - 2311 vol.5
    Save to Project icon | Click to expandQuick Abstract | PDF file iconPDF (422 KB) |  | HTML iconHTML  

    Popular secure public key cryptosystems such as RSA and Diffie-Hellman are based on hard problems like factorization and discrete logarithms. These systems often require large prime numbers the size of 300 decimal digits long for the systems to be secure. The generation of large prime numbers is difficult, and larger prime numbers will be required as advances in parallel computing makes factorization of large numbers faster. In this paper, algorithms for digital signatures and public key cryptosystems using multilayer perceptrons (MLPs) are proposed. The security of the algorithm is based on the difficult problem of solving non-linear simultaneous equations. Instead of needing large prime numbers, the algorithm requires multiple real numbers that can be easily generated. View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • The SOM-TSP method for the three-dimension city location problem

    Page(s): 2552 - 2555 vol.5
    Save to Project icon | Click to expandQuick Abstract | PDF file iconPDF (385 KB) |  | HTML iconHTML  

    Up to now, shortening the installation time of electronic parts by chip-mounter has been researched by using the SOM-TSP method. This research aims to examine the effectiveness of the optimization ability of the SOM-TSP method when a multi-dimensional city is located. View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • On detection of confused blood samples using self-organizing maps and genetic algorithm

    Page(s): 2233 - 2238 vol.5
    Save to Project icon | Click to expandQuick Abstract | PDF file iconPDF (696 KB) |  | HTML iconHTML  

    A SOM (self-organizing map)-based detection of confusion of blood test data referred to as CBC (complete blood count) data is proposed. Firstly, the method based on only SOM is shown. The learning data applied to SOMs are generated by subtracting the immediately anterior CBC data of subjects from the present CBC data. All the neurons in the second layer of SOM trained by applying the above learning data are roughly divided into the following two clusters: a cluster with neurons reacting to regular input data, and a cluster reacting to irregular input data which are generated by subtraction between confused CBC data. So, if the firing neuron belongs to the latter cluster, it is presumed that the confusion arises among CBC data of some subjects. Next, a method based on both SOM and GA (genetic algorithm) is shown. With the exception of selecting some elements, which instruct the weights to be updated in the second layer of CBC data by means of GA, the learning and the detection strategy adopted by this method are similar to those by the firstly proposed method. Experimental results on detecting the confusion, which arises among CBC data of 750 subjects, show that the second proposed method produces the second layer which achieves the high accuracy of detection especially when the input data, not to be employed during the learning, are applied. View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • Classifying brain shapes

    Page(s): 2678 - 2682 vol.5
    Save to Project icon | Click to expandQuick Abstract | PDF file iconPDF (708 KB) |  | HTML iconHTML  

    Brain shapes do not necessarily form a continuum in some descriptor space, but may form clusters related to pre-determined genetical factors or acquired diseases. As a feasibility study for introducing a suitable descriptor space, the use of modal analysis was tested on a large brain database acquired in healthy young subjects. Significant shape differences due to gender were found, and intra-gender clusters were determined. View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • Self organizing fuzzy neural network: an application to character recognition

    Page(s): 2640 - 2644 vol.5
    Save to Project icon | Click to expandQuick Abstract | PDF file iconPDF (509 KB) |  | HTML iconHTML  

    Character recognition is a very important field in DSP. Many different methods are used for this purpose. The ANN technique based on back propagation algorithm is very slow as its computational complexity is very high. On the other hand the Self-orthogonal ANN offers less computational complexity but it is not able to deal with the uncertainty associated with the input data sequence. Hence, fuzzy logic is applied in this case. The fuzzy logic based self-orthogonal neural network has been applied to the scale changed and distorted characters only. The problem of invariance to rotation has been discussed using the four layered feed forward fuzzy neural network. View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • A stochastic nonlinear evolution model of neuronal activity with random amplitude

    Page(s): 2497 - 2501 vol.5
    Save to Project icon | Click to expandQuick Abstract | PDF file iconPDF (470 KB) |  | HTML iconHTML  

    In this paper we propose a new stochastic nonlinear evolution model with the stochastic amplitude in neuronal activities to obtain the average number density, which is used to describe the neurocommunication among population of neurons. The average number density is a function of the amplitude, phase and time. The number density of the diffusion process of neurocommunication is given for the active states of populations of coupled oscillators under perturbation by both periodic stimulation and random noise. Particularly, the evolution model presented in this paper can be used to describe the stochastic evolution process of the amplitudes in activities of multiple interactive neurons. View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • Supervised map ICA: applications to brain functional MRI

    Page(s): 2259 - 2263 vol.5
    Save to Project icon | Click to expandQuick Abstract | PDF file iconPDF (698 KB) |  | HTML iconHTML  

    This paper gives a method to control or organize itself an activation pattern of fMRI maps obtained by ICA (independent component analysis). The presented method uses an additional term to the convex divergence's gradient. The following merits are observed: (i) Prior knowledge can be effectively used so that obtained activation patterns properly reflect the task on the subject. (ii) Difficulty of finding the appropriate activation pattern due to the permutation can be avoided. Experiments on brain fMRI maps for visual cortices are tried and reported. View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • Adaptive neural network ensemble that learns from imperfect supervisor

    Page(s): 2561 - 2565 vol.5
    Save to Project icon | Click to expandQuick Abstract | PDF file iconPDF (457 KB) |  | HTML iconHTML  

    In training supervised-type neural networks, the quality of the training data is one of the most important factors in deciding the quality of the neural networks. Unfortunately, in real world problems, error-free training data are not always easy to obtain. For complex data, it is always possible that erroneous training samples are included, causing to decrease the performance of the neural networks. In this research, we propose a model of neural network ensemble that, through a competition mechanism, has an ability to automatically train one of its members to learn only from the correct training patterns, thus minimizing the effect of the imperfect data. View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • A design method of DNA chips using hierarchical self-organizing maps

    Page(s): 2244 - 2248 vol.5
    Save to Project icon | Click to expandQuick Abstract | PDF file iconPDF (541 KB) |  | HTML iconHTML  

    We introduce a design method of DNA chips using self organizing maps (SOM) and hierarchical self-organizing maps (H-SOM). DNA chips are powerful tools for sequencings and SNP (single nucleotide polymorphism) analyses of DNA sequences. A DNA chip is an array of DNA probes which can be hybridized with complement subsequences in the target sequence. However, conventional DNA chips are showing tendency to be comprised of large number of long probes and get large in size to achieve high resolution. To shrink the size of DNA chips, design method is considered to be important. To solve this problem, we applied SOM to extract common features of DNA sequences using proper number of probes which efficiently cover the target sequence with sufficient resolutions. Furthermore, H-SOM can design the chip comprised of long probes more efficiently than SOM. We have designed some DNA chips from the sequence data in genome database using our SOM and H-SOM algorithm and evaluated the chips by computer simulations of SNP analyses. View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • Sign language recognition using sensor gloves

    Page(s): 2204 - 2206 vol.5
    Save to Project icon | Click to expandQuick Abstract | PDF file iconPDF (352 KB) |  | HTML iconHTML  

    This paper examines the possibility of recognizing sign language gestures using sensor gloves. Previously sensor gloves are used in games or in applications with custom gestures. This paper explores their use in Sign Language recognition. This is done by implementing a project called "Talking Hands", and studying the results. The project uses a sensor glove to capture the signs of American Sign Language performed by a user and translates them into sentences of English language. Artificial neural networks are used to recognize the sensor values coming from the sensor glove. These values are then categorized in 24 alphabets of English language and two punctuation symbols introduced by the author. So, mute people can write complete sentences using this application. View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • Fuzzy classification based identification of voltage sag via wavelets

    Page(s): 2381 - 2385 vol.5
    Save to Project icon | Click to expandQuick Abstract | PDF file iconPDF (542 KB) |  | HTML iconHTML  

    Increasing awareness of power quality issues, deregulation, use of consumer devices sensitive to power system disturbance and possibility of making up some of the inherent design limitations through monitoring based operational strategies have created a need for extensive monitoring of the power system operation. Voltage disturbance is a common phenomenon in electric power distribution system operation. A fuzzy diagnostic procedure is proposed for detecting cause of voltage disturbance, so that appropriate remedial procedures could be initiated during system operation. The method uses indices like PN factor, characteristic voltage, and zero sequence voltage and also proposes an index termed frequency jump index, extracted from zero sequence voltage using wavelets. View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • Hebbian learning in an automatic gender identification by speech system

    Page(s): 2409 - 2413 vol.5
    Save to Project icon | Click to expandQuick Abstract | PDF file iconPDF (404 KB) |  | HTML iconHTML  

    This work presents an Automatic Gender Identification (AGI) algorithm based on Eigenfiltering. A Maximum Eigenfilter is implemented by means of an Artificial Neural Network (ANN) trained via Generalized Hebbian Learning (GHL). The Eigenfilter uses Principal Component Analysis (PCA) to perform maximum information extraction from the speech signal, which enhances correlated information and improves the pattern analysis. Also, a well known speech processing technique is applied, the Mel-Frequency Cepstral Coefficients (MFCC). This technique is a classical approach for speech feature extraction, and it is a very efficient way to represent physiological voice parameters. The pattern classification uses a Radial Basis Function (RBF) ANN. Experimental results have shown that the identification algorithm overall performance was widely increased by the Eigenfiltering process. View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • Australian All Ordinaries Index: re-examine the utilities of the explanatory variables using three different measures

    Page(s): 2335 - 2339 vol.5
    Save to Project icon | Click to expandQuick Abstract | PDF file iconPDF (512 KB) |  | HTML iconHTML  

    The stock markets are generally nonlinear dynamic systems. Therefore, estimating stock market output depends mainly on nonlinear relationships of input variables. Additionally researchers have demonstrated that financial markets display self-similarity. To forecast such systems, a nonlinear modeling tool is required. The paper compares accuracy of prediction using the following techniques: neural network, linear regression and exponential moving average, and measures their performance using mean absolute percentage error (MAPE), Thiel's U Statistics (U-STAT) and R Square (RSQ). Using the All Ordinaries Index and a sliding window through the data, our results show that the explanatory variables can improve the predictive power of two techniques when predicting future changes in the index. Only the MAPE but not the other two measures show the effect in this setting. View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • An encoding technique based on word importance for the clustering of Web documents

    Page(s): 2207 - 2211 vol.5
    Save to Project icon | Click to expandQuick Abstract | PDF file iconPDF (494 KB) |  | HTML iconHTML  

    We present a word encoding and clustering technique that groups Web documents based on the importance of the words that appear in the documents. We use a two level self-organizing map architecture to generate clusters of words and documents. We propose that by capturing word importance information of words, similar documents can be then clustered to assist in Web document retrieval. A Web document retrieval system is presented to demonstrate how this approach could. be integrated into Web search. View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • Multi-modular architecture based on convolutional neural networks for online handwritten character recognition

    Page(s): 2444 - 2448 vol.5
    Save to Project icon | Click to expandQuick Abstract | PDF file iconPDF (506 KB) |  | HTML iconHTML  

    In this paper, several convolutional neural network architectures are investigated for online isolated handwritten character recognition (Latin alphabet). Two main architectures have been developed and optimised. The first one, a TDNN, processes online features extracted from the character. The second one, a SDNN, relies on the off-line bitmaps reconstructed from the trajectory of the pen. Moreover, an hybrid architecture called SDTDNN has been derived, it allows the combination of on-line and off-line recognisers. Such a combination seems to be very promising to enhance the character recognition rate. This type of shared weights neural networks introduces the notion of receptive field, local extraction and it allows to restrain the number of free parameters in opposition to classic techniques such as multi-layer perceptron. Results on UNIPEN and IRONOFF databases for online recognition are reported, while the MNIST database has been used for the off-line classifier. View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • Neural network algorithm for solving ray-tracing problem

    Page(s): 2665 - 2668 vol.5
    Save to Project icon | Click to expandQuick Abstract | PDF file iconPDF (419 KB) |  | HTML iconHTML  

    This work is dedicated to the study of neural network method for solving of ray-tracing task, which appears in 3D visualization algorithms. Physical representation of the task is the problem of finding the nearest point of the "vision" ray crossing with the surfaces of the scene. Application: Real time 3D visualization, rendering of the complex scenes, containing semitransparent, reflecting, diffusive objects, soft shadows and volume light sources. View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • Nonlinear Phillips curves in the Euro area and USA? Evidence from linear and neural networks models

    Page(s): 2142 - 2146 vol.5
    Save to Project icon | Click to expandQuick Abstract | PDF file iconPDF (511 KB) |  | HTML iconHTML  

    This paper applies neural network methodology to inflation forecasting in the Euro-area and the USA. Neural network methodology outperforms linear forecasting methods for the Euro Area at forecast horizons of one, three, and six month horizons, while the linear model is preferable for US data. The nonlinear estimation shows that unemployment is a significant predictor of inflation for the Euro Area. Neither model detects a significant effect of unemployment on inflation for the US data. View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • Artificial neural network approach in determining voltage stability in power system networks

    Page(s): 2304 - 2307 vol.5
    Save to Project icon | Click to expandQuick Abstract | PDF file iconPDF (405 KB) |  | HTML iconHTML  

    Voltage stability problems have been one of the major concerns for electric utilities as a result of system heavy loading. As electric power systems are operated under increasingly stressed conditions, the ability to maintain voltage stability becomes a growing concern. This paper reports on an investigation on the application of artificial neural networks (ANNs) in voltage stability assessment. A multilayer feedforward ANN with error back propagation learning is proposed for calculation of voltage stability index (L). Extensive testing of the proposed ANN based approach indicates its viability for power system voltage assessment. Test results are presented on two sample power systems. View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • Dynamical scaling in on-line hand-written characters' matching

    Page(s): 2449 - 2453 vol.5
    Save to Project icon | Click to expandQuick Abstract | PDF file iconPDF (447 KB) |  | HTML iconHTML  

    On-line signature verification and/or recognition of handwritten characters are becoming more and more important in the network society. When these problems are considered, one fundamental method will be to find the corresponding points between the test one and the template. For this purpose, we proposed a matching method using DP (dynamic programming) matching and dynamical scaling parameters. This paper supplements the scaling aspect and the numerical examples that will support the validity of this method. View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • Estimation of inner information representations in time series prediction and bi-directionalization effect of computing architecture

    Page(s): 2147 - 2151 vol.5
    Save to Project icon | Click to expandQuick Abstract | PDF file iconPDF (541 KB) |  | HTML iconHTML  

    A bi-directional computing architecture for time series prediction, which computes not only the future prediction transformation but also the past prediction one, is proposed recently and applied to several prediction tasks. According to the previous studies, an improvement of the prediction performances has been observed with different kinds of data sets. Nevertheless, its detailed mechanism for temporal signal processing is not clear yet. Then, in order to solve this problem, the model's responses are investigated based on the principal component analysis approach in this paper. As a result, it is found experimentally that an enrichment of the inner information representations gives the model an advantage on signal processing abilities through bi-directionalization of the computing architecture. View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.