Zarita Zainuddin

Modified wavelet neural network in function approximation and its application in prediction of time-series pollution data

Properly designing a wavelet neural network (WNN) is crucial for achieving the optimal generalization performance. In this paper, in order to improve the predictive capability of WNNs, the types of activation functions used in the hidden layer of the WNN were varied. The modified WNNs were then applied in approximating a benchmark piecewise function. Subsequently, performance comparisons with other developed methods in studying the same benchmark function were made. An assessment analysis showed that this proposed approach outperformed the rest. The efficiency of the modified WNNs was explored through a real-world application problem-specifically, the prediction of time-series pollution data at Texas of United States. The comparative experimental results showed that integrating different wavelet families into the hidden layer of WNNs leads to superior performance.

Optimisation of soda pulping variables for preparation of dissolving pulps from oil palm fibre.

Summary A water prehydrolysis-soda pulping sequence for the preparation of dissolving pulps from oil palm fibre (empty fruit bunches) was investigated using a response surface methodology (RSM) statistical experiment design. Five response variables of screened yield, Kappa number, α-cellulose, viscosity and ash content were statistically analyzed with respect to the three input variables of pulping temperature (T), time-at-temperature (t) and alkali level (A), while keeping the prehydrolysis conditions constant. Optimum conditions were: T = 161°C, t = 100 min and A= 26.1%. Values predicted by RSM for screened yield, Kappa number, α-cellulose, viscosity and ash content at the optimum are 31.2%, 6.0, 96.9%, 16.1 cps and 0.15%, respectively. These values were experimentally verified and very close agreement between experimental and predicted values was obtained.

Reliable multiclass cancer classification of microarray gene expression profiles using an improved wavelet neural network

Abstract Properly designing a wavelet neural network (WNN) is crucial for achieving the optimal generalization performance. In this paper, two different approaches were proposed for improving the predictive capability of WNNs. First, the types of activation functions used in the hidden layer of the WNN were varied. Second, the proposed enhanced fuzzy c-means clustering algorithm—specifically, the modified point symmetry-based fuzzy c-means (MSFCM) algorithm—was employed in selecting the locations of the translation vectors of the WNN. The modified WNN was then applied to heterogeneous cancer classification using four different microarray benchmark datasets. The comparative experimental results showed that the proposed methodology achieved an almost 100% classification accuracy in multiclass cancer prediction, leading to superior performance with respect to other clustering algorithms. Subsequently, performance comparisons with other classifiers were made. An assessment analysis showed that this proposed approach outperformed most of the other classifiers.

Reliable epileptic seizure detection using an improved wavelet neural network.

BACKGROUND Electroencephalogram (EEG) signal analysis is indispensable in epilepsy diagnosis as it offers valuable insights for locating the abnormal distortions in the brain wave. However, visual interpretation of the massive amounts of EEG signals is time-consuming, and there is often inconsistent judgment between experts. AIMS This study proposes a novel and reliable seizure detection system, where the statistical features extracted from the discrete wavelet transform are used in conjunction with an improved wavelet neural network (WNN) to identify the occurrence of seizures. METHOD Experimental simulations were carried out on a well-known publicly available dataset, which was kindly provided by the Epilepsy Center, University of Bonn, Germany. The normal and epileptic EEG signals were first pre-processed using the discrete wavelet transform. Subsequently, a set of statistical features was extracted to train a WNNs-based classifier. RESULTS The study has two key findings. First, simulation results showed that the proposed improved WNNs-based classifier gave excellent predictive ability, where an overall classification accuracy of 98.87% was obtained. Second, by using the 10th and 90th percentiles of the absolute values of the wavelet coefficients, a better set of EEG features can be identified from the data, as the outliers are removed before any further downstream analysis. CONCLUSION The obtained high prediction accuracy demonstrated the feasibility of the proposed seizure detection scheme. It suggested the prospective implementation of the proposed method in developing a real time automated epileptic diagnostic system with fast and accurate response that could assist neurologists in the decision making process.

Modification of the Decision-Making Capability in the Social Force Model for the Evacuation Process

The Social Force Model is one of the most successful microscopic pedestrian models that represent the well-organized phenomena of the pedestrian flow. The model has been modified for evacuation process by incorporating physical forces when contact exists, on one hand, and incorporating factors into the preferred velocity to govern the individuals behavior corresponding to the situation under consideration (normal or evacuation) on the other hand. The latter incorporation has enhanced the ability of the model to represent the decision-making process of pedestrians. However, the variety of pedestrians abilities to make decisions in emergency situations has not been incorporated properly into the model. In this article we enhance the decision-making capability of the independent pedestrians first by improving the assessment process of selecting an exit from the set of exits available in the physical environment by considering a new factor (crowd at exits); and second, by incorporating following crowds as a new feature for those who are independent. A simulation of an emergency situation inside a room is performed to validate our work.

Wavelet neural networks applied to pulping of oil palm fronds

In the organosolv pulping of the oil palm fronds, the influence of the operational variables of the pulping reactor (viz. cooking temperature and time, ethanol and NaOH concentration) on the properties of the resulting pulp (yield and kappa number) and paper sheets (tensile index and tear index) was investigated using a wavelet neural network model. The experimental results with error less than 0.0965 (in terms of MSE) were produced, and were then compared with those obtained from the response surface methodology. Performance assessment indicated that the neural network model possessed superior predictive ability than the polynomial model, since a very close agreement between the experimental and the predicted values was obtained.

Double approximate identity neural networks universal approximation in real lebesgue spaces

The aim of this study is to investigate that some classes of feedforward neural networks exist such that they have universal approximation property. Based on the double approximate identity a main theorem is presented. The result shows the universal approximation capability of double approximate identity neural networks in real two dimensional compact Lebesgue integrable subspaces.

On the Use of Wavelet Neural Networks in the Task of Epileptic Seizure Detection from Electroencephalography Signals

Abstract This paper investigates the feasibility and effectiveness of wavelet neural networks (WNNs) in the task of epileptic seizure detection. The electroencephalography (EEG) signals were first pre-processed using discrete wavelet transforms (DWTs). This was followed by the feature selection stage, where two sets of four representative summary statistics were computed. The features obtained were fed into the input layer of WNNs. Three different activation functions were used in the hidden nodes of WNNs – Gaussian, Mexican Hat, and Morlet wavelets. A 10-fold cross validation was performed and the performance assessment revealed that the proposed classifiers achieved high overall classification accuracy, which showed the prominence of WNNs in this binary classification task. The best combination to be used was the WNNs that employed Morlet wavelet as the activation function, with Daubechies wavelet of order 4 in the feature extraction stage. The cross comparison done showed that the classification accuracy achieved by WNNs was comparable to those of other artificial intelligence-based classifiers. It was also demonstrated that a classifier would perform better if input features with higher dissimilarity index were used.

On the Universal Approximation Capability of Flexible Approximate Identity Neural Networks

This study presents some class of feedforward neural networks to investigate the universal approximation capability of continuous flexible functions. Based on the flexible approximate identity, some theorems are constructed. The results are provided to demonstrate the universal approximation capability of flexible approximate identity neural networks to any continuous flexible function.

Improved wavelet neural network for early diagnosis of cancer patients using microarray gene expression data

In clinical practice, diagnostic dilemmas are frequently encountered in discriminating the heterogeneous cancers into distinct types. This paper reports an improved machine learning approach based on the wavelet neural network (WNN), which associates a feature selection method, namely, the conditional T-test. It is used in the development of cancer classification by using benchmark microarray data. The experimental results showed that the proposed classifiers achieved a superior accuracy, which ranges from 92% to 100%. Performance comparisons are also made with other classifiers which show that this proposed approach outperforms most of them.