Tung Kai Xu ¹, M. P. Paulraj ¹

Affiliations
1 School of Mechatronic Engineering, Universiti Malaysia Perlis, Perlis, MY

Abstract

Aggressiveness is one of the most important human characteristics that enable humans to achieve and reach to a higher level in their day to day activities. Conventionally, the aggressiveness of a subject is normally measured using Buss Perry aggressiveness questionnaire method. The validity of the aggressiveness score of a subject measured using this questionnaire method highly depends on the honesty of the subject while giving feedback for the questionnaire. Further, the aggressive level of a subject will change with respect to time and other environmental factors. Considering the variability of aggressive level, two simple methods, namely, task based mean of inter channel correlation coefficient method and inter trial task based channel correlation coefficient method have been proposed to estimate the aggressiveness of a subject while playing a smart phone game. A simple protocol to measure the EEG signals from the subjects while playing a smart phone game is proposed and EEG signals from 10 different subjects are obtained. Using the developed methods, the Buss Perry Aggressive Index, BPAI values were computed and analyzed together with the conventional Buss Perry Questionnaire based aggressive level index values. From the results it has been observed that subjects with higher BPAI value will get into higher aggressiveness state quickly and recover back to relaxing state quickly compare to subjects with lower BPAI value. For subject with BPAI level near to the group classifying limit might possess both group characteristic of Net Aggressiveness Index, NAI development when expose to induction of aggressiveness. Therefore the proposed task based mean of inter channel correlation coefficient method and inter trial task based channel correlation coefficient method can be used to measure the aggressive index values.

Keywords

Aggressiveness Level Index, Channel Correlation, Correlation Coefficient, EEG, Smart Phone Game.

Full Text

   

M. A. Yusnita ¹, M. P. Paulraj ², Sazali Yaacob ³, R. Yusuf ¹, M. Nor Fadzilah ¹

Affiliations
1 Faculty of Electrical Engineering, Universiti Teknologi MARA Malaysia, Permatang Pauh - 13500, Penang, MY
2 School of Mechatronic Engineering, University Malaysia Perlis, Ulu Pauh - 02600, Perlis, MY
3 Universiti Kuala Lumpur Malaysian Spanish Institute, Kulim Hitech Park, Kulim - 09000, Kedah, MY

Abstract

To date, Malaysian English (MalE) accents arises from different ethnics of its populace are scarcely investigated using empirical methods that give a decisive conclusion to treat MalE as either uniform or non-uniform variety. The popularly used Mel-Frequency Cepstral Coefficients (MFCC) and Linear Prediction Coefficients (LPC) as feature extractors fail to perform well under noisy conditions. This paper proposes two new methods and noise less-susceptible feature extractors to mitigate the deficiency of MFCC and LPC. Statistical descriptors of Mel-Bands Spectral Energy (MBSE) is an enhancement of traditional filter-bank analysis, however, increases fourfold as much the feature size. This issue is tackled by proposing a transformation using principle component analysis to generate a new PCA-MBSE feature set. Experimental results indicated promising accuracy rates of 92.7% and 93.0% using the proposed PCA-MBSE features to recognize between the Malay, Chinese and Indian accents of MalE speech for the male and female datasets respectively. It was found that under severe noisy conditions, the standard MFCC and LPC features started to deteriorate faster than the MBSE-based features. PCA-MBSE features showed the most robust quality where its performance was just slightly deteriorated by 17.1% and 13.6% as compared to MBSE features i.e. 33.1% and 31.3% for the male and female datasets respectively. Further poor results of LPC features were obtained indicating deterioration rates of 40.2% and 32.7%, while that of MFCC features of 35.7% and 36.8% for the male and female datasets respectively.

Keywords

Accent Recognition, K-Nearest Neighbors, Linear Prediction Coefficients, Malaysian English, Mel-Bands Spectral Energy, Mel-Frequency Cepstral Coefficients, Principle Component Analysis

Full Text

   

Sathees Kumar Nataraj ¹, M. P. Paulraj ¹, Sazali Bin Yaacob ¹, Abdul Hamid Adom ¹

Affiliations
1 School of Mechatronic Engineering, Universiti Malaysia Perlis, Perlis, MY

Abstract

This paper describes the classification of navigational tasks to command a navigation system incorporated with a communication device using thought and visually evoked potentials. To develop a navigation system with communication aid for the neuromuscular disorder community, simple protocol using TEP and VEP responses has been introduced in this research work. The developed protocol has seven basic tasks such as forward, left, right, yes, no, help and relax; these basic seven task are used to control the wheel chair navigation system and also perform voice communication using an oddball paradigm. The proposed system records the brain wave signals using a wireless EEG amplifier from ten subjects while the subjects were imagining and visualizing the seven different visual tasks. For each subject, the recorded brain wave signals are pre-processed to extract the six Electroencephalography rhythmic activities and segmented into frames of equal samples. Then, this study presents the higher order spectra based features to categorize the TEP and VEP tasks using bispectrum estimation algorithm. Further, statistical features such as the mean and entropy of the bispectral magnitude are extracted and formed as a feature set. To develop a customized classification system for individual responses, the extracted feature sets are classified using Multi layer neural networks and from the results it is observed that the entropy of bispectral magnitude feature using VEP based NN model has the maximum classification accuracy of 99.29% and the mean of bispectral magnitude feature using TEP based NN model has the minimum classification accuracy of 72.14%.

Keywords

Bispectrum Estimation, Customized-Intelligent Robot Chair with Communication Aid, Multi Layer Neural Network, Thought Evoked Potentials

Full Text

   

M. A. Yusnita ¹, M. P. Paulraj ², Sazali Yaacob ³, R. Yusuf ¹, M. Nor Fadzilah ¹

Affiliations
1 Faculty of Electrical Engineering, Universiti Teknologi MARA Malaysia, Permatang Pauh - 13500, Penang, MY
2 School of Mechatronic Engineering, University Malaysia Perlis, Ulu Pauh - 02600, Perlis, MY
3 Universiti Kuala Lumpur Malaysian Spanish Institute, Kulim Hitech Park, Kulim - 09000, Kedah, MY

Abstract

The standard speech feature extractors such as Mel-Frequency Cepstral Coefficients (MFCC) and Linear Prediction Coefficients (LPC) fail to perform well under noisy conditions. In this paper two noise less-susceptible features are proposed to mitigate the deficiency of MFCC and LPC. Statistical descriptors of Mel-Bands Spectral Energy (MBSE) is applied to the traditional filter-bank analysis, however, this technique increases the feature size. This issue is tackled by proposing a transformation using principle component analysis to generate a new PCA-MBSE feature set. Two types of utterances namely isolated words and continuous speech were elicited from 103 university volunteers in a controlled room to collect speech signals from three main ethnic groups in Malaysia. This study employed two classifiers namely K-nearest neighbors and artificial neural networks to recognize between the Malay, Chinese and Indian accents. Experimental results using independent test samples technique indicated promising accuracy rates of 92.7% and 93.0% using the proposed PCA-MBSE features to recognize between the Malay, Chinese and Indian accents on the male and female datasets respectively. It was found that under severe noisy conditions, the standard MFCC and LPC features started to deteriorate faster than the MBSE-based features. PCA-MBSE features showed the most robust quality where its performance was just slightly deteriorated by 17.1% and 13.6% as compared to MBSE features i.e. 33.1% and 31.3% on the male and female datasets respectively. Further poor results of LPC features were obtained indicating deterioration rates of 40.2% and 32.7%, while that of MFCC features of 35.7% and 36.8% for the male and female datasets respectively. As a conclusion, Malaysian English is a not a uniform English variety colored by its diverse ethnic nuances. Incorporating accent analyzers using the proposed techniques in automatic speech recognition can contribute a substantial improvement in noisy environment.

Keywords

Accent Recognition, K-Nearest Neighbors, Linear Prediction Coefficients, Malaysian English, Mel-Bands Spectral Energy, Mel-Frequency Cepstral Coefficients, Principle Component Analysis

Full Text

   

Jackie Teh ¹, M. P. Paulraj ¹

Affiliations
1 School of Mechatronic Engineering, University Malaysia Perlis, Perlis, MY

Abstract

A brain-actuated wheelchair can be used to aid the movement of differentially enabled communities who face much difficulties while commuting from one place to another. In this research work, the active brain signals emanated from subjects while performing four different kinesthetic motor imagery tasks are recorded using Electroencephalography (EEG). Three different feature sets, namely, Fractal Dimension (FD), Mel-Frequency Cepstral Coefficients (MFCCs) and combined features of FD with MFCCs are extracted from the recorded EEG signals. The extracted features are then associated to classify the type of motor imagery tasks and three feedforward multi-layer Perceptrons trained with Levenberg-Marquardt method are developed. The performance of the three features are evaluated in term of classification rate and compared. Simple Elman network and NARX network models are then developed using the extracted features and evaluated. From the results, it is observed that the Elman network model trained with combined features of FD with MFCCs has yielded a higher classification accuracy for all the 5 subjects in the range of 98.98-100percent. The obtained result clearly indicates that the Elman network and combined features of FD with MFCCs has potential to classify the four different motor imagery tasks.

Keywords

Brain Computer Interface, Elman Neural Network, Feedforward Multi-Layered Perceptron Neural Network, Fractal Dimension, Mel-Frequency Cepstral Coefficients, Nonlinear Autoregressive Exogenous Model, Recurrent Neural Network

Full Text

   

N. I. E. Farhana ¹, M. S. Abdul Majid ¹, M. P. Paulraj ¹, E. Ahmadhilmi ²

Affiliations
1 School of Mechatronic Engineering, University Malaysia Perlis, Arau, MY
2 Advanced Material Research Centre (AMREC), SIRIM Berhad, Kulim, MY

Abstract

Currently, the volume fraction of a glass fibre/matrix based composite material is being assessed only by destructive techniques. Instead of changing or destroying the structure, a new non-destructive approach based on vibration technique is proposed in this research. Further, the main objective of this paper is on the determination of fibre/matrix volume fractions using vibration analysis. A complete experimental protocol has been developed to record the vibration signals produced from experimental plates with different volume fractions and thicknesses. The recorded vibration signals were analyzed both in time and frequency domains. Subsequently, statistical parameter features from each thickness was extracted and associated to the volume fraction levels. Artificial Neural Network (ANN) models were then developed to classify the level of volume fraction. The classification performance of the developed network models were in the range of 80-98 percent. From the results, it has been observed that the network model with frequency band based features has yielded a better classification performance. This proves that the method implemented can be used as the alternatives to the ASTM D2584−11 for determination of volume fraction of a glass fibre/matrix composite plate using vibration analysis.

Keywords

Composite, Feed-Forward Neural Network, Non-Destructive Testing, Statistical Features, Vibration Signal, Volume Fraction.

Full Text

   

Eric Tiong Kung Woo ¹, M. P. Paulraj ¹, Abdul Hamid Adom ¹

Affiliations
1 School of Mechatronic Engineering, University Malaysia Perlis, Pauh, MY

Abstract

Electroencephalography (EEG) signal is a biological signal which can be associated to the mental task of a person. A Brain-Computer Interface (BCI) can be designed such that the mental activity accounted for the visual perception of a person can be recorded and subsequently converted into a control signal for controlling the movement of a wheelchair. Comparison was made between the Multi-Layered Perceptron feed-forward network (MLP) and Nonlinear Autoregressive Exogenous model (NARX) as a variant of Recurrent Neural Network (RNN). The networks were designed to discriminate the different brain activities when the subject was being presented with different visual stimuli. The trained network models have yielded an average accuracy of 93.3% for MLP models and 98.1% for NARX models.

Keywords

Brain Computer Interface (BCI), Electroencephalography (EEG), Levenberg-Marquardt Training Algorithm (LM), Multi-Layered Perceptron Neural Network (MLP), Nonlinear Autoregressive Exogenous Model (NARX), Power Spectral Density (PSD), Visual Perception.

Full Text