Prima Dewi Purnamasari

EEG based patient emotion monitoring using relative wavelet energy feature and Back Propagation Neural Network

In EEG-based emotion recognition, feature extraction is as important as the classification algorithm. A good choice of features results in higher recognition rate. However, there is no standard method for feature extraction in EEG-based emotion recognition, especially for real time monitoring, where speed of computation is crucial. In this work, we assess the use of relative wavelet energy as features and Back Propagation Neural Network (BPNN) as classifier for emotion recognition. This method was implemented in simulated real time emotion recognition by using a publicly accessible database. The results showed that relative wavelet energy and BPNN achieved an average recognition rate of 92.03%. The highest average recognition rate was achieved when the time window was 30s.

Development of Filtered Bispectrum for EEG Signal Feature Extraction in Automatic Emotion Recognition Using Artificial Neural Networks

The development of automatic emotion detection systems has recently gained significant attention due to the growing possibility of their implementation in several applications, including affective computing and various fields within biomedical engineering. Use of the electroencephalograph (EEG) signal is preferred over facial expression, as people cannot control the EEG signal generated by their brain; the EEG ensures a stronger reliability in the psychological signal. However, because of its uniqueness between individuals and its vulnerability to noise, use of EEG signals can be rather complicated. In this paper, we propose a methodology to conduct EEG-based emotion recognition by using a filtered bispectrum as the feature extraction subsystem and an artificial neural network (ANN) as the classifier. The bispectrum is theoretically superior to the power spectrum because it can identify phase coupling between the nonlinear process components of the EEG signal. In the feature extraction process, to extract the information contained in the bispectrum matrices, a 3D pyramid filter is used for sampling and quantifying the bispectrum value. Experiment results show that the mean percentage of the bispectrum value from 5 × 5 non-overlapped 3D pyramid filters produces the highest recognition rate. We found that reducing the number of EEG channels down to only eight in the frontal area of the brain does not significantly affect the recognition rate, and the number of data samples used in the training process is then increased to improve the recognition rate of the system. We have also utilized a probabilistic neural network (PNN) as another classifier and compared its recognition rate with that of the back-propagation neural network (BPNN), and the results show that the PNN produces a comparable recognition rate and lower computational costs. Our research shows that the extracted bispectrum values of an EEG signal using 3D filtering as a feature extraction method is suitable for use in an EEG-based emotion recognition system.

Artificial Neural Networks Based Emotion Classification System through Relative Wavelet Energy of EEG Signal

Emotion classification from EEG brain signal has been a widely research topic recently, because of the complexity of processing the multi channels of the brain signal and also the problem on mapping the human emotion itself. This paper discusses the technique to determine the emotion classification from EEG brain signal using a relative wavelet energy (RWE) as a feature vector and an artificial neural networks (ANN) as a classifier. In this research, two types of ANN classifier was utilized and analyzed, namely, Back-propagation Neural Networks (BPNN) and Probabilistic Neural Networks (PNN). Also reducing the number of the EEG channel to be processed is investigated, in order to decrease the computational cost of the classification system. Results showed that the recognition rate of the reduced utilized channels up to 4 channel are incomparable with that of the full utilization of 32 channels. However, 8 and 14 channels still produced sufficient recognition rate. It is also confirmed from experiments that the BPNN shown as a more reliable classifier compare with the PNN method.

3D object implementation on bicycling at ui virtual reality application based on 3D-Gamestudio

This paper reviews 3D computer technology and our experience in creating a virtual bicycling environment at the University of Indonesia Green Eco-campus. We explain the implementation of the VR environment using 3D-Games Studio and our experience in viewing the result with VR device, i.e. 3D E-Dimensional wireless goggle.

Cloud computing network design for high performance computing implementation on openstack platform

This paper presents networking design strategies in cloud infrastructure built using Openstack platform and explore scalability issues related to its infrastructure. Experiments discussed in this paper are conducted to study the performance of Openstack network based on the implementation of Neutron in order to provide recommendations for optimal network design suitable for high performance computing. Network performance parameters such as throughput, packet loss and latency were evaluated based on TCP and UDP data transmission using IPerf benchmarking tool. The results show that in general, Openstack did not exhibit network bandwidth bottlenecks. However, the location where virtual machine was instantiated and how the network was segmented will affect network performance. This paper also discussed the traffic flow used to analyze the network performances of different virtual machines network topology in the cloud and provide an optimal cloud networking design and recommendations that yields to a better high-performance computing platform.

Performance comparison of text-based sentiment analysis using recurrent neural network and convolutional neural network

One biggest challenge in sentiment analysis is that it should include Natural Language Processing (NLP), to make the machine understand the human language. With the current development of Artificial Neural Network (ANN), with its implementation, computer can learn to understand human language by such learning mechanism There are many types of ANN and for this research Convolutional Neural Network (CNN) and Recurrent Neural Network (RNN) were used and compared on their performance. The text data for the sentiment analysis was taken from Stanford publication and transformation from text to vectors were conducted using word2vec. The result shows that RNN is better than CNN. Even the difference of accuracy is not significant with 88.35% ± 0.07 for RNN and 87.11% ± 0.50 for CNN, the training time for RNN only need 8.256 seconds while CNN need 544.366 seconds.

Automatic essay grading system with latent semantic analysis and learning vector quantization

Automatic essay grading system called SIMPLE-O (Sistem Penilai Esai Otomatis) that developed by Department of Electrical Engineering, Faculty of Engineering, Universitas Indonesia was built using PHP. This system was developed for helping lecturer assessing the examination, especially with essay form. Currently, the SIMPLE-O is still developed using C programming language for implementing more methods in development that can only be done in that language to improve the performance of the system. To increase the accuracy, Learning Vector Quantization (LVQ) algorithm is implemented in the development due to its ability for supervised classification. The number of data samples in LVQ training phase are affecting the essay scoring performance, less data used will lead to decrease in the result accuracy of the validation phase. Moreover, singular value that generated by Frobenius norm and vector angle pre-processing will also affect the scoring accuracy. But, the number of words-per-column when creating the LSA matrix did not have any significant effect. At the end, SIMPLE-O with LVQ has an average accuracy of 53.57%, 41.66% higher than the system that did not use LVQ. This accuracy performance was still low due to the missing of the words similarity feature. In the previous version of SIMPLE-O, this feature can improve the performance of the essay grading system significantly.

Parallel Processing Design of Latent Semantic Analysis Based Essay Grading System with OpenMP

In this paper, a parallel processing design for an short-answer essay grading system based on Latent Semantic Analysis (LSA) is presented. This system was developed to ease the lecturers for short-essay assessment process. With the number of documents to be assessed in an exam period, efficiency of processing and time are needed. Parallel processing is a solution for advancing the system to be implemented for wide scope of use. The design was implemented using OpenMP Application Programming Interface (API) to achieve this research purpose. Most of the systems processes utilize the outer loop or repetitive process with data level parallelism to process the same process for different data at a time. Early experiment on SVD process has promising result for parallelization implementation compared to the serial process. With parallelization, SVD could process 50 documents in 1.773 microseconds (ms), 3 times faster than the serial process in 5.676 ms. There were 2 other experiments to obtain the optimum number of matrix size and number of threads for parallelization; the experiment showed that two threads division on SVD process is the best option for this system and more threads are appropriate for more complex implementation and larger LSA based system.

Cross-Language Plagiarism Detection System Using Latent Semantic Analysis and Learning Vector Quantization

Computerized cross-language plagiarism detection has recently become essential. With the scarcity of scientific publications in Bahasa Indonesia, many Indonesian authors frequently consult publications in English in order to boost the quantity of scientific publications in Bahasa Indonesia (which is currently rising). Due to the syntax disparity between Bahasa Indonesia and English, most of the existing methods for automated cross-language plagiarism detection do not provide satisfactory results. This paper analyses the probability of developing Latent Semantic Analysis (LSA) for a computerized cross-language plagiarism detector for two languages with different syntax. To improve performance, various alterations in LSA are suggested. By using a linear vector quantization (LVQ) classifier in the LSA and taking into account the Frobenius norm, output has reached up to 65.98% in accuracy. The results of the experiments showed that the best accuracy achieved is 87% with a document size of 6 words, and the document definition size must be kept below 10 words in order to maintain high accuracy. Additionally, based on experimental results, this paper suggests utilizing the frequency occurrence method as opposed to the binary method for the term–document matrix construction.

Relative wavelet bispectrum feature for alcoholic EEG signal classification using artificial neural network

This paper proposes a novel relative wavelet bispectrum (RWB) approach for EEG signal feature extraction method to differentiate the signal between the alcoholic over the non-alcoholic subjects. Firstly, the EEG signal is calculated for its autocorrelation frequencies as the basic step in the bispectrum calculation. Then, the discrete wavelet transform (DWT) is applied substituting the FFT which usually is used in the bispectrum calculation. Lastly, the relative value of each frequency band is calculated for both the approximation and the details parts, producing the RWB. The proposed methodology is implemented in an alcoholic automated detection system using 1200 data samples from UCI EEG Database for alcoholism. Based on the experiments, the setting value of lag in the autocorrelation calculation was evidently very influential on the recognition rate obtained, i.e. the maximum value for the lag was the best. Using cross validation, the highest results from RWB feature extraction method with ANN classifier achieved about 90% recognition rate.