Neural Network Pattern Recognition Research Articles

Cognitive Decision Making in Multiple Sensor Monitoring of Robot Assisted Polishing

A multiple sensor monitoring system, comprising acoustic emission, strain and voltage sensors, was utilised during an experimental campaign of robot assisted polishing of steel bars for on-line evaluation of workpiece surface roughness. Two feature extraction procedures, based on conventional statistics and wavelet packet transform algorithms, were applied to the detected sensor signals in order to extract features to be fed to cognitive methods based on neural network pattern recognition paradigms seeking for correlations with the surface roughness of the polished workpiece.

Automated detection of Epilepsy using Wavelet Features

Epilepsy is generally considered as a group of neurological disorders characterized by epileptic seizures. It is often confirmed with an electroencephalogram (EEG). But identification of epilepsy has to be done by skilled neurologist. This paper proposes an efficient methodology for automatic detection of ictal and healthy EEG signals which is the ultimate goal of machine learning, which has performed efficient classification. We used discrete wavelet transform for feature extraction and obtained wavelet coefficients. Neural network pattern recognition tool is used for classification. The performance of the proposed method is evaluated using in terms of sensitivity, specifity and accuracy.

Application of support vector machine for pattern classification of active thermometry-based pipeline scour monitoring

Pipeline scour monitoring is becoming one of the key requirements in oil and gas industry. To implement scour monitoring for offshore pipeline, a monitoring system that based on active thermometry is proposed. Our previous investigations have shown that the system has provided many advantages over traditional scour monitoring methods. In this paper, a novel scour automatic detection scheme based on nonlinear curve fitting and support vector machine (SVM) is proposed to realize automatic diagnosis of pipeline scour. On account of the varied heat transfer patterns of a line heat source in sediment and water scenarios, the experimental temperature profiles are nonlinearly fitted to their theoretical models. Features extracted by nonlinear curve fitting can dramatically reduce the dimensions of the data. Subsequently, the extracted features are inputted into SVM classifier to judge where the pipeline is exposed to water or buried in the sediment. In order to evaluate the performance of SVM, SVM with different kernel functions are compared with the back-propagation neural networks, which is the most popular neural network for pattern recognition and classification. Results show that the SVM model with radial basis function kernel outperformed other classification models. Finally, aiming to obtain the optimal heating time of the system, the optimal SVM model is employed to recognize datasets with different heating time. Copyright © 2014 John Wiley & Sons, Ltd.

ACCELERATED BEACH EROSION ALONG THE UPPER TEXAS COAST

The upper Texas coast has been severely eroded and beach erosion is going on. It is not extreme comparing past one but its strength is still strong by a magnitude. There are some methods and numbers to determine stability of beach by historical beach observation but they are evaluated only for a certain location. It may not be available in the places which have rare observation data such as the upper Texas coast. The study focused on factors to determine the stability of beach against beach erosion. The pattern recognition of neural network was applied to classification of the factors such as beach width, dune height from field observed data.

Improving pattern recognition accuracy of partial discharges by new data preprocessing methods

In this paper, raw data of partial discharges (PDs) in solid, oil, and air insulation materials are measured experimentally in a high voltage laboratory for 18 samples. Then, three new methods for preprocessing the data based on first, second, and infinite signal norms and besides autocorrelation function (ACF) are proposed. Eventually, feed-forward back propagation (FFBP), radial basic function (RBF) neural networks, and neural network pattern recognition toolbox (nprtool) are used to recognize the patterns of the processed data. The results of the new methods are compared with phase resolved partial discharge (PRPD) method which is common in previous studies. Thanks to the new preprocessing methods, correlation factor in FFBP network, error value in RBF network, and classification percentage in nprtool become 0.9867, 0.0001 and 96.4%, respectively. Moreover, it is concluded that PDs process is a stationary random process which can be estimated by Gauss–Markov process.

Detection of multi‐occupancy using device‐free passive localisation

Indoor device-free passive localisation (DfPL) technology uses a received signal strength indication (RSSI)-based method to record variances of a measured signal where a person being tracked is not carrying any electronic device that can be used to estimate the location. The system monitors the changes in the RSSI measurements caused by the presence of a human body in an indoor environment. For example, it is known that the resonance frequency of water is 2.4 GHz and the human body contains >70% water. Thus, the human body attenuates the wireless signal reacting as an absorber. Wireless communication signal strengths between a number of nodes, using IEEE 802.11 or 802.15.4 standards, show that communication links covering distinct areas cannot be affected simultaneously by only one person. Thus, the authors have deployed a novel system that can identify multi-occupants in an environment using patterns of motion from those monitored areas. A pattern recognition neural network was used to identify two people in the environment. No other work based on the DfPL technique has focused on multi-occupancy.

RETRACTED: Hybrid Hidden Markov Models and Neural Networks Based on Face Recognition

Retracted paper: In this paper, the basic principles of HMM, HMM studied three major issues need to be addressed as well as overflow problems in the practical application of how to solve the HMM. Because artificial neural network (ANN) with anti-noise, adaptive, learning ability, recognition speed, etc., taking into account the characteristics of the common features of speech recognition and pattern recognition and artificial neural networks have, this article will get a mixed combination of HMM in ANN model, using ANN to make up for some deficiencies of HMM. Experiments show that the hybrid model recognition rate than the HMM model increased by 4%, but the algorithm still has many defects to be resolved.

Detection and Classification of Single and Combined Power Quality Disturbances Using Neural Networks

The detection and classification of power quality (PQ) disturbances have become a pressing concern due to the increasing number of disturbing loads connected to the power line and the susceptibility of certain loads to the presence of these disturbances; moreover, they can appear simultaneously since, in any real power system, there are multiple sources of different disturbances. In this paper, a new dual neural-network-based methodology to detect and classify single and combined PQ disturbances is proposed, consisting, on the one hand, of an adaptive linear network for harmonic and interharmonic estimation that allows computing the root-mean-square voltage and total harmonic distortion indices. With these indices, it is possible to detect and classify sags, swells, outages, and harmonics-interharmonics. On the other hand, a feedforward neural network for pattern recognition using the horizontal and vertical histograms of a specific voltage waveform can classify spikes, notching, flicker, and oscillatory transients. The combination of the aforementioned neural networks allows the detection and classification of all the aforementioned disturbances even when they appear simultaneously. An experiment under real operating conditions is carried out in order to test the proposed methodology.

Prediction of Successful Surgery Outcome in Lumbar Disc Herniation Based on Artificial Neural Networks

Introduction The aim of this study, was to develop an artificial neural networks (ANNs) model for predicting successful surgery outcome in lumbar disc herniation (LDH). Materials and Methods An ANN model and a logistic regression (LR) model were used to predict outcomes. The age, gender, duration of symptoms, smoking status, surgical level, visual analog scale of leg/back pain, the Zung depression scale, and the Japanese Orthopaedic Association score, were determined as the input variables for the established ANN model. The Macnab classification was used for outcome assessment. ANNs on data from LDH patients, who had surgery, were trained to predict 2-year successful discectomy using several input variables. Sensitivity analysis to the established ANN model was used to identify the relevant variables. For evaluating the two models, the area under a receiver operating characteristic curve, accuracy rate of predicting, and Hosmer-Lemeshow (H-L) statistics were considered. Results A total of 203 (96 male, 107 female, mean age 48.3 ± 9.8 years) patients were categorized into training, testing, and validation datasets consisting of 101, 51, and 51 cases, respectively. Surgical successful outcome was categorized as: excellent, 32.0%; good, 40.9%; fair, 20.7%; and poor, 6.4% at 2-year follow-up. Compared with the LR model, the ANN model showed better results: accuracy rate, 95.8%; H-L statistic, 41.5%; and AUC, 0.82% of patients, respectively. Conclusion The findings show that ANNs can predict successful surgery outcome with a high level of accuracy in LDH patients. Such information is of use in the clinical decision-making process. Disclosure of Interest None declared References Fardon DF, Milette PC; Combined Task Forces of the North American Spine Society, American Society of Spine Radiology, and American Society of Neuroradiology. Nomenclature and classification of lumbar disc pathology. Recommendations of the Combined task Forces of the North American Spine Society, American Society of Spine Radiology, and American Society of Neuroradiology. Spine 2001;26(5):E93-E113 Li YC, Liu L, Chiu WT, Jian WS. Neural network modeling for surgical decisions on traumatic brain injury patients. Int J Med Inform 2PO.069;57(1):1–9 Bishop CM. Neural Networks for Pattern Recognition. Oxford, UK: Clarendon; 1995 Azimi P, Benzel EC, Shahzadi S, Azhari S, Mohammadi HR. Use of artificial neural networks to predict surgical satisfaction in patients with lumbar spinal canal stenosis. J Neurosurg Spine 2014;20(3):300–305 Price DD, McGrath PA, Rafii A, Buckingham B. The validation of visual analogue scales as ratio scale measures for chronic and experimental pain. Pain 1983;17(1):45–56 Azimi P, Mohammadi HR, Montazeri A. An outcome measure of functionality and pain in patients with lumbar disc herniation: a validation study of the Japanese Orthopedic Association (JOA) score. J Orthop Sci 2012;17(4):341–345 Zung WW, Richards CB, Short MJ. Self-rating depression scale in an outpatient clinic. Further validation of the SDS. Arch Gen Psychiatry 1965;13(6):508–515 Macnab I. Chapter 14. Pain and disability in degenerative disc disease. Clin Neurosurg 1973;20:193–196 Cross SS, Harrison RF, Kennedy RL. Introduction to neural networks. Lancet 1995;346(8982):1075–1079 Rughani AI, Dumont TM, Lu Z, et al. Use of an artificial neural network to predict head injury outcome. J Neurosurg 2010;113(3):585–590 Finneson BE. A lumbar disc surgery predictive score card. Spine 1978;3(2):186–188

Construction of Artificial Neural Networks for Pattern Recognition Using a Successive Geometric Segmentation Method

This work presents a new method for artificial neural networks construction and training based on successive geometric segmentation (SGSM). This methodology is capable of generating both the network topology and neurons weights without parameters’ specification. The SGSM groups the data of each class into hyper-boxes (HBs) aligned in accordance with the largest axis of its points’ distribution. If the HBs are linearly separable, then a separating hyperplane may be identified resulting in a neuron. If it is not, then a segmentation technique is applied to divide the data into smaller classes for new HBs. For each subdivision, new neurons are added to the network. The tests show a rapid method with a high success rate.

Design and Comparative Analysis of SVM and PR Neural Network for Classification of Brain Tumour in MRI

grams (EEGs) or MRI are progressively emerging as a significant measure of brain activity and they possess immense potential for the diagnosis and treatment of mental and brain diseases and abnormalities. This research paper presents an automated system for efficient classification of brain tumours in MRI images using Support Vector Machine (SVM). It works on the principle of a vector which supports in training of classifier using train images. The results obtained from the above method are compared with those obtained from Artificial Neural Networks (ANNs). This method uses pattern recognition algorithms for the classification of tumour in MR images. First, a pattern recognition neural network is created and then trained. The trained pattern recognition neural network when fed with a test MR image, effectively classifies the images. The performance analysis shows that qualitative results obtained from the proposed model are comparable with those obtained by ANN. The experimental results demonstrate the effectiveness of the proposed system in classification of brain tumour in MR images. The implementation is done using MATLAB R2012b using Artificial Neural Network and Image Processing Toolbox.

Construction of Artificial Neural Networks for Pattern Recognition Using a Successive Geometric Segmentation Method

Construction of Artificial Neural Networks for Pattern Recognition Using a Successive Geometric Segmentation Method

Study on Artificial Neural Networks to Identify Sedimentary Microfacies

Based on a number of stratigraphic sedimentary information included in log data, application of the Artificial Neural Network to identify sedimentary microfacies from well logging data can complete the series auto-interpreting. The application can improve the auto-interpreting accuracy and make us get more satisfied results. Ten parameters from the well logging curves are selected for to describing their shape characteristics when the deposition patterns of 8 in gas-bearing formation of Upper Paleozoic group, Ordos basin are studied. Effective parameters were selected on the basis of cores, and based on artificial neural network pattern recognition technique; the sedimentary microfacies of well cross section were auto-interpreted. About 300 wells and the results were interpreted by using the software. The software will be fit for the researchers who have the experiences of geological interpretation and some backgrounds of local geology.

Complicated Graphics Model based on Neural Network of Particle Swarm

The past decades have seen the great progress pattern recognition and image understanding, motivated by a wide range of real world applications. The previous approaches mostly based on the feature extraction and recognition methods, and usually suffer from the problem of weak discrimination power. In this paper, we propose a complex model based on particle swarm and neural network pattern recognition methods. The proposed approach uses the class label of the input sample as the actual output of the neural network which has the maximum node corresponds to the class. To evaluate the effectiveness of the proposed approach, we experimentally compare the particle swarm neural network based approach with and SVM classification algorithm. The experimental results show that, both the recognition accuracy and training time of the particle swarm and neural network based approach, are better than that of SVM. The particle swarm neural network pattern recognition in complex graphics showed a great advantage in application.

Automatic seizure detection based on Teager Energy Cepstrum and pattern recognition neural networks

About 1–3% of the world population suffers from epilepsy. A long term inpatient/ambulatory electroencephalogram (EEG), lasting from a few hours to several days, which definitely contains hallmarks of epilepsy, is required clinically to diagnose, monitor and localize the epileptogenic zone. The traditional method relies on well-trained neurophysiologists who visually inspect the entire lengthy EEG signals, which is tedious and time-consuming. Therefore, many automated epileptic detection systems have been developed and such systems significantly reduce the time taken to review off-line the long-term EEG recordings and facilitate the neurologist to diagnose and treat more patients in a given time. There are not many studies that have explored, to a sufficient depth, the features used in other domains of signal processing, for example, the Teager energy cepstrum (TECEP), attempted use in seizure detection. Epileptic seizures are abnormal sudden discharges in the brain with signatures manifesting in the EEG recordings by frequency changes and increased amplitudes. These changes, in this work, are captured through static and dynamic features derived from TECEP. We compared the performance of the baseline TECEP and its two composite vectors with those of the traditional cepstrum (CEP) and its corresponding two composite vectors, in EEG epileptic seizure detection. The first composite vector includes velocity cepstral coefficients and the second includes velocity, as well as, acceleration cepstral coefficients. The comparison is tried on eight different classification problems which encompass all the possible discriminations in the medical field related to epilepsy, using pattern recognition neural network (PRNN). In this case, it is found that the overall performance of both the Teager energy composite vectors excels those of traditional cepstral composite vectors. The static and dynamic features derived from TECEP outperform those derived from CEP, suggesting their suitability in epilepsy seizure detection.

Comparison between BP and RBF Neural Network Pattern Recognition Process Applied in the Droplet Analyzer

Among various pattern recognition methods used for liquid identification, the method based on neural network has the advantages of robustness and fault tolerance, which can study and adapt to the uncertain system. The waveform analysis is exploited for feature extraction of the liquid droplet fingerprint (LDF) in this paper, and the liquid identification is carried out by means of BP and RBF neural network. The experimental results proved that the recognition rate is excellent in both of these two methods. In condition that the training data is limited, RBF network is better than BP network in recognition speed and rate.

Pattern Recognition Neural Networks in Classification Groups of Partial Removable Dentures Lifetime

Classification is one of the most common research work associated with neural networks. This paper cast lighton the use of this technique as a robust statistical procedure for modeling data regarding problem of classification groups of partial removable dentures lifetime. Groups are either 1 which refers to lifetime exceeds average, and 0 which refers to lifetime less than or equal average lifetime. The main purpose of this work is to provide a new technique of classification as well as to stimulate further research interests and efforts in this topic.

Image Segmentation with Artificial Neural Networs Alongwith Updated Jseg Algorithm

Image segmentation plays a very crucial role in computer vision. Computational methods based on JSEG algorithm is used to provide the classification and characterization along with artificial neural networks for pattern recognition. It is possible to run simulations and carry out analyses of the performance of JSEG image segmentation algorithm and artificial neural networks in terms of computational time. A Simulink is created in Matlab software using Neural Network toolbox in order to study the performance of the system. In this paper, four windows of size 9*9, 17*17, 33*33 and 65*65 has been used. Then the corresponding performance of these windows is compared with ANN in terms of their computational time. Image segmentation is a process of separating images into small section that can be used as meaningful objects in computing and searching techniques. These segments are developed by using a number of processing techniques that are clustering, J-fractal image calculation. In Bottom Up approach techniques like JSEG, pixels of image are processed and divided into classes. These classes are used for indentifying image segments, once pixels are labeled using classes. The classes are clustered to divide image pixels and for image segments. Existing JSEG image segmentation technique uses K-Means for clustering. But there are a number of clustering variations of K-means available which can be used for improving results of image segmentation. K means clustering algorithm suffers from problem of over segmentation which is a big problem in process of segmentation in case of JSEG. We will be experimenting with available clustering techniques for building image segmentation. Many of existing segmentation techniques such as direct clustering methods in color space work well on homogeneous color regions. Natural scenes are rich in color and texture. Parameter estimation is a difficult problem and requires good homogeneous region for robust estimation. A new approach called JSEG is proposed towards this goal. This approach does not tend to estimate a specific model for a texture region. It test for the homogeneity of a given color pattern, which is computationally more feasible method. JSEG algorithm can be improved by adding enhanced classification and modified K means clustering.

Thin Film Coated QCM-Sensors and Pattern Recognition Methods for Discrimination of VOCs

Abstract discrimination and quantification of volatile organic compounds (VOCs) using a non-selective sensor requires a combination of sensors followed by pattern recognition methods. Based on this concept, this paper deals with the discrimination of gas from the responses of several gas sensors coated with different type of polymer. Quartz crystal microbalance (QCM) electrodes were coated from hexamethyldisiloxane (HMDSO), hexamethyldisilazane (HMDSN) and tetraethoxysilane (TEOS) for the elaboration of gas sensors with different chemical affinity towards VOC molecules. The sensitivity of the elaborated QCM-based sensors was evaluated by monitoring the frequency shifts of the quartz exposed to different concentrations of volatile organic compounds, such as; ethanol, benzene and chloroform. The sensors responses data have been used for the identification and quantification of VOCs. The principal component analysis (PCA) and the neural-network (NNs) pattern recognition analysis were used for the discrimination of gas species and concentrations. Good separation among gases has been obtained using the principal component analysis. The feed-forward multilayer neural network (MLNNs) with a hidden layer and trained by Broyden Fletcher Goldfarb Shanno (BFGS) Quasi Newton algorithm has been implemented in order to identify and quantify the VOCs. By increasing the number of the neuron in the hidden layer, the precision of the estimate concentration increases. The approach is standard, however its application on the elaborated sensors have not been studied in depth so far.

Label free detection of pseudorabies virus infection in Vero cells using laser force analysis

The rapid and robust identification of viral infections has broad implications for a number of fields, including medicine, biotechnology and biodefense. Most detection systems rely on specific molecules, such as nucleic acids or proteins, to identify the target(s) of interest. These molecules afford great specificity, but are often expensive, labor-intensive, labile and limited in scope. Label free detection methods seek to overcome these limitations by instead using detection methods that rely on intrinsic properties as a basis for identifying and separating species of interest and thus do not rely on specific prior knowledge of the target. Optical chromatography, one such technique, uses the balance between optical and fluidic drag forces within a microfluidic channel to determine the optical force on cells or particles. Here we present the application of individual optical force measurements as a means of investigating pseudorabies virus infection in Vero cells. Optical force differences are seen between cells from uninfected and infected populations at a multiplicity of infection as low as 0.001 and as soon as 2 hours post infection, demonstrating the potential of this technique as a means of detecting viral infection. Through the application of a pattern recognition neural network, individual cell size data are combined with optical force as a means of classifying cell populations. Potential applications include the early detection of bloodborne pathogens for the prevention of sepsis and other diseases as well as the detection of biological threat agents.