Djamal, Esmeralda Contesa
Universitas Jenderal Achmad Yani

Published : 16 Documents Claim Missing Document
Claim Missing Document
Check
Articles

Found 16 Documents
Search

Identification of Speed and Unique Letter of Handwriting Using Wavelet and Neural Networks Djamal, Esmeralda C.; Febriyanti, Febriyanti
Proceeding of the Electrical Engineering Computer Science and Informatics Vol 2: EECSI 2015
Publisher : IAES Indonesia Section

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.11591/eecsi.2.543

Abstract

Graphology is scientific method to evaluation personality and emotion condition through handwriting and signature. There are many features to identify personality so that previous researches made handwriting analysis automatically. There are page margins, spacing, baseline, vertical zone, font size, slant, pen pressure, and the type t letter. While other studies used features of signature. As image, the analysis of graphology is divided into two approaches that graphics features and segmentation digit each character. This research integrated both of approach to identify personality of handwriting. It used speed and the type of a, d, i m, t letters as features using structure analysis and artificial neural networks. Type of letter recognition was done after character segmentation. Wavelet transform was used to improve recognition. The proposed methods could be used to identify personality of handwriting. Identification of speed feature using structure analysis toward page margin, spacing between lines, and spacing between words that gave 81% accuracy. While identification of unique letters using neural network with multilayer perceptron architecture, which gave 74% accuracy. Variations training data greatly affect recognition.
Classification of Motor Imagery and Synchronization of Post-Stroke Patient EEG Signal Fadiyah, Arifah Ummul; Djamal, Esmeralda C.
Proceeding of the Electrical Engineering Computer Science and Informatics Vol 6: EECSI 2019
Publisher : IAES Indonesia Section

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.11591/eecsi.v6.1935

Abstract

Stroke attacks often cause disability, so the need for rehabilitation to restore patient's motor skills. Electroencephalogram (EEG) is an instrument that can capture electrical activity in the brain. Some post-stroke patients have brain electrical dysfunction so that EEG signal can achieve such as amplitude decrease, and wave differences from symmetric channels. However, EEG signal analysis is not easy because it has high complexity and small amplitude. However, information from EEG signals is beneficial, including for stroke identification. This study proposes the identification of EEG signals from post-stroke patients using wavelet extraction and Backpropagation Levernberg-Marquardt. EEG signals are recorded, extracted imagery motor variables, and synchronization of symmetric channels. The results of the study provide that the accuracy for identifying post-stroke EEG signals is 100% for training data and 79.69 % for new data. Research also shows that the use of learning rates affects accuracy. The smaller the learning rate provided accuracy is better. However, it had consequences for computing time so that the optimal learning rate is 0.0001.
Emotion and Attention of Neuromarketing Using Wavelet and Recurrent Neural Networks Ar Rasyid, Muhammad Fauzan; Djamal, Esmeralda C.
Proceeding of the Electrical Engineering Computer Science and Informatics Vol 6: EECSI 2019
Publisher : IAES Indonesia Section

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.11591/eecsi.v6.1939

Abstract

One method concerning evaluating video ads is neuromarketing. This information comes from the viewer's mind, thus minimizing subjectivity. Besides, neuromarketing can overcome the difficulties of respondents who sometimes do not know the response to the video ads they watch. Neuromarketing is based on neuropsychology, which is sourced from the human brain through electrical activity signals recorded by Electroencephalogram. Usually, Neuropsychology consists of emotions, attention, and concentration. This research proposed the Wavelet method and Recurrent Neural Networks to measure the emotional and attention variable of neuropsychology in real-time every two seconds while watching video ads. The results showed that Wavelet and Recurrent Neural Networks could provide training data accuracy of 100% and 89.73% for new data. The experiment also gave that the RMSprop optimization model for the weight correction contributed to higher correctness of 1.34% than the Adam model. Meanwhile, using Wavelet for extraction can increase accuracy by 4%.
Speaker and Speech Recognition Using Hierarchy Support Vector Machine and Backpropagation F. Fadlilah, Asti; C. Djamal, Esmeralda
Proceeding of the Electrical Engineering Computer Science and Informatics Vol 6: EECSI 2019
Publisher : IAES Indonesia Section

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.11591/eecsi.v6.1969

Abstract

Voice signal processing has been proposed to improve effectiveness and facilitate the public, such as Smart Home. This study aims a smart home simulation model to move doors, TVs, and lights from voice instructions. Sound signals are processed using Mel-frequency Cepstrum Coefficients (MFCC) to perform feature extraction. Then, the voice is recognized by the speaker using a hierarchy Support Vector Machine (SVM). So that unregistered speakers are not processed or are declared not having access rights. For the process of recognizing spoken words such as "Open the Door”,"Close the Door","Turn on the TV","Turn off the TV","Turn on the Lights" and "Turn Offthe Lights" are done using Backpropagation. The results showed that hierarchy SVM provided an accuracy of 71% compared to the single SVM of 45%.
Paraphrase Detection Using Manhattan's Recurrent Neural Networks and Long Short-Term Memory Aziz, Achmad; Djamal, Esmeralda Contessa; Ilyas, Ridwan
Proceeding of the Electrical Engineering Computer Science and Informatics Vol 6: EECSI 2019
Publisher : IAES Indonesia Section

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.11591/eecsi.v6.1973

Abstract

Natural Language Processing (NLP) is a part of artificial intelligence that can extract sentence structures from natural language. Some discussions about NLP are widely used, such as Recurrent Neural Networks (RNN) and Long Short-Term Memory (LSTM) to summarize papers with many sentences in them. Siamese Similarity is a term that applies repetitive twin network architecture to machine learning for sentence similarity. This architecture is also called Manhattan LSTM, which can be applied to the case of detecting paraphrase sentences. The paraphrase sentence must be recognized by machine learning first. Word2vec is used to convert sentences to vectors so they can be recognized in machine learning. This research has developed paraphrase sentence detection using Siamese Similarity with word2vec embedding. The experimental results showed that the amount of training data is dominant to the new data compared to the number of times and the variation in training data. Obtained data accuracy, 800,000 pairs provide accuracy reaching 99% of training data and 82.4% of new data. These results are better than the accuracy of the new data, with half of the training data only yielding 64%. While the amount of training data did not effect on training data.
Detection of EEG Signal Post-Stroke Using FFT and Convolutional Neural Network Djamal, Esmeralda C.; Furi, Widiyanti Isni; Nugraha, Fikri
Proceeding of the Electrical Engineering Computer Science and Informatics Vol 6: EECSI 2019
Publisher : IAES Indonesia Section

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.11591/eecsi.v6.2013

Abstract

Stroke is a condition that occurs when the blood supply to the brain is disrupted or reduced. It may be caused by a blockage (ischemic stroke) or rupture of a blood vessel (hemorrhagic stroke) so that it can cause disability. Therefore patients need to undergo rehabilitation. One of the procedures of monitoring of the recovery of stroke patients using the National Institutes of Health Stroke Scale (NIHSS) method, but sometimes subjectively. Electroencephalogram (EEG) is an instrument that can measure electrical activity in the brain, including abnormalities caused by stroke. This study investigates EEG signal detection in post-stroke patients using Fast Fourier Transform (FFT) and 1D Convolutional Neural Network (1D CNN). Fast Fourier Transform (FFT) extraction can increase accuracy from 60% to 80.3% from the use of Adam's optimization model. Meanwhile, the AdaDelta model gave 20% accuracy without FFT. And its condition increased to 79.9% with FFT extraction. Therefore, Adam's stability has the advantage of remembering to use hyper-parameter. On the other hand, FFT is beneficial for directing information used for the use of 1D CNN, thus increasing accuracy. The results showed that using of Fast Fourier Transform (FFT) in identification could increase accuracy by 45-80% compared to identification using only 1D CNN. Meanwhile, the results of the study show that the relative weight correction model using Adaptive Moment Estimation (Adam) provided higher accuracy compared to the Adaptive learning rate (AdaDelta).
Identification of Speed and Unique Letter of Handwriting Using Wavelet and Neural Networks Contessa Djamal, Esmeralda
Proceeding of the Electrical Engineering Computer Science and Informatics Vol 2: EECSI 2015
Publisher : IAES Indonesia Section

Show Abstract | Download Original | Original Source | Check in Google Scholar | Full PDF (948.163 KB) | DOI: 10.11591/eecsi.v2.493

Abstract

Handwriting  stroke  reflects  the  personality  and emotional    condition.    Graphology    is    scientific    method    to evaluation  personality  through  handwriting.  There  are  many features  in  graphology  to  identify  personality.  Several  previous researches  used  page  margins,  spacing,  baseline,  vertical  zone, font  size,  and  the  type  of  unique  letter  t.  Other  research  also identify the personality of signatures. This research uses feature writing  speed  and  the  type  of  letters  a,  d,  i  m,  and  t  to  identify personalities   using   structural   analysis   and   artificial   neural networks.  To  improve  accuracy,  image  writing  extracted  using wavelet transform. The system is built with the approach of the structure  and  symbol  has  been  implemented  in  software.  The results show a unique type of letter recognition by 74%, and the speed  feature  by  60%  recognition.  Variations  training  data greatly affect recognition results.
Spoken Word Recognition Using MFCC and Learning Vector Quantization Djamal, Esmeralda C.; Nurhamidah, Neneng; Ilyas, Ridwan
Proceeding of the Electrical Engineering Computer Science and Informatics Vol 4: EECSI 2017
Publisher : IAES Indonesia Section

Show Abstract | Download Original | Original Source | Check in Google Scholar | Full PDF (398.748 KB) | DOI: 10.11591/eecsi.v4.1043

Abstract

Identification of spoken word(s) can be used to control external device. This research was result word identification in speech using Mel-Frequency Cepstrum Coefficients (MFCC) and Learning Vector Quantization (LVQ). The output of system operated the computer in certain genre song appropriate with the identified word. Identification was divided into three classes contain words such as "Klasik", "Dangdut" and "Pop", which are used to playing three types of accordingly songs. The voice signal is extracted by using MFCC and then identified using LVQ. The training and test set were obtained from six subjects and 10 times trial of the words "Klasik", "Dangdut" and "Pop" separately. Then the recorded sound signal is pre-processed using Histogram Equalization, DC Removal and Pre-emphasize to reduce noise from the sound signal, and then extracted using MFCC. The frequency spectrum generated from MFCC was identified using LVQ after passing through the training process first. Accuracy of the testing results is 92% for identification of training sets while testing new data recorded using different SNR obtained an accuracy of 46%. However, the test results of new data recorded using the same SNR with training data has an accuracy of 75.5%.
EEG Based Emotion Monitoring Using Wavelet and Learning Vector Quantization C. Djamal, Esmeralda; Lodaya, Poppi
Proceeding of the Electrical Engineering Computer Science and Informatics Vol 4: EECSI 2017
Publisher : IAES Indonesia Section

Show Abstract | Download Original | Original Source | Check in Google Scholar | Full PDF (1077.524 KB) | DOI: 10.11591/eecsi.v4.1053

Abstract

Emotional identification is necessary for example in Brain Computer Interface (BCI) application and when emotional therapy and medical rehabilitation take place. Some emotional states can be characterized in the frequency of EEG signal, such excited, relax and sad. The signal extracted in certain frequency useful to distinguish the three emotional state. The classification of the EEG signal in real time depends on extraction methods to increase class distinction, and identification methods with fast computing. This paper proposed human emotion monitoring in real time using Wavelet and Learning Vector Quantization (LVQ). The process was done before the machine learning using training data from the 10 subjects, 10 trial, 3 classes and 16 segments (equal to 480 sets of data). Each data set processed in 10 seconds and extracted into Alpha, Beta, and Theta waves using Wavelet. Then they become input for the identification system using LVQ three emotional state that is excited, relax, and sad. The results showed that by using wavelet we can improve the accuracy of 72% to 87% and number of training data variation increased the accuracy. The system was integrated with wireless EEG to monitor emotion state in real time with change each 10 seconds. It takes 0.44 second, was not significant toward 10 seconds.
Optimalisasi Distribusi Harga Tiket Pesawat berdasarkan Kepadatan Rute Menggunakan Algoritma Genetika Novianti, Sri Hutamy; Djamal, Esmeralda C.; Komarudin, Agus
Jurnal Teknik Informatika dan Sistem Informasi Vol 5 No 2 (2019): JuTISI
Publisher : Maranatha University Press

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.28932/jutisi.v5i2.1756

Abstract

The development of the aviation industry in Indonesia in the past decade has risen sharply. One of the impacts of the development of the aviation industry was the presence of a multilevel tariff concept. Where, the concept is the variation in ticket prices in one class with slightly different facilities such as the difference in penalty fees for making refunds and rebooking. The concept of multilevel rates is usually referred to as sub-class rates. One application of the sub-class tariffs in economic classes is divided into four types of sub-classes special promo sub-classes, promo sub-classes, then affordable sub-class and flexible sub-class. One optimization method of getting a combination that meets the requirements without having to try all possibilities is the Genetic Algorithm. The chromosomes built represent 10 subclasses on 9 routes so that they have 90 genes. The use of genetic algorithms originated from the generation of an initial population of 8 chromosomes with a length of 90 genes performed randomly, evaluation of the compatibility function was then selected using the Rank based fitness technique, crosses using Multi-Point Crossover, mutations with the Mutation Insertion technique. The system built was tested with two conditions each of eight tests with 100 generations. First, the test uses the mutation method of three subclass codes on four routes at a capacity of 150 seats, obtained the largest match value of Rp. 750,752,200 and the smallest Rp. 662,283,100. And testing with the mutation method of three subclass codes on eight routes of 150 seat capacity obtained the largest match value of Rp. 763,265,300 and the smallest Rp. 547,396,200. The results of testing the mutation method on eight routes resulted in a higher match value compared to the mutation method on four routes. The system has been implemented in software so that it can provide recommendations on the number of ticket passes distributed in the economic subclass.