Jurnal Ilmu Komputer dan Informasi
Not a member yet
247 research outputs found
Sort by
Embedded Deep Learning System for Classification of Car Make and Model
Full text linkAutomatic car make, and model classification is essential to support activities of intelligent traffic systems in urban areas, such as surveillance, traffic information collection, statistics, etc. In order to classify this data, we need an embedded system approach for real-time car recognition. Many approaches could be made, from image processing to machine learning. Recently, the development of the Convolutional Neural Network has spurred various research in the Area. ResNet, Inception, DenseNet, and NasNet are some of the most commonly used Neural Network based method that is used to classify images. In this research, these Neural Network methods are going to be compared in classifying vehicle make and model in the Stanford dataset. The dataset contains 196 different labels. Several evaluation metrics are used to compare the performance of the methods. From the experiment, the InceptionV3 method achieved the best performance of the AUROC ratio for training the dataset under 50 epochs. Other methods that achieve a high AUROC value tends to have a higher computational time. Real-time simulations have shown that the embedded system is capable of classifying a 100 % success rate for six concurrent users
Design and Development of EcoSense: Android-Based Incentivized Environmental Campaign App
Full text linkThis research paper presents the design and development of EcoSense, an Android-based incentivized environmental campaign app, using the Design Science Research Methodology (DSRM). The DSRM approach involved six steps: problem identification, objective definition, design and development, demonstration, and evaluation. The problem of insufficient participation in environmental activities in Indonesia was identified using the Five Whys method, and user requirements were collected through an online survey. Various design artifacts were created to ensure the app is user-friendly, and the app was evaluated based on user engagement and participation. The results suggested that the app has the potential to effectively engage users in environmental campaigns, as evidenced by the relatively high retention and conversion rates. The study has some limitations, such as a small sample size and limited evaluation metrics, but future research can address these limitations and expand on the findings
Weld Defect Detection and Classification based on Deep Learning Method: A Review
Full text linkThe inspection of weld defects utilizing nondestructive testing techniques based on radiography is essential for ensuring the operability and safety of weld joints in metals or other materials. During the process of welding, weld defects such as cracks, cavity or porosity, lack of penetration, slag inclusion, and metallic inclusion may occur. Due to the limitations of manual interpretation and evaluation, recent research has focused on the automation of weld defect detection and classification from radiographic images. The application of deep learning algorithms enables automated inspection. The deep learning architectures for building weld defect classification models were discussed. This paper concludes with a discussion of the achievements of automation methods and a presentation of the research recommendations for the future
Optimization of 2D-CNN Setting for the classification of covid disease using Lung CT Scan
Full text linkRT-PCR is considered the best diagnostic tool. Previous studies have demonstrated the reliability of CNN in classifying classifications, but CNN requires a lot of training data. Meanwhile, at the CT Scan clinic, patients are limited. Therefore, exploration of 2D-CNN settings is proposed to optimize CNN performance on limited data. We compare: (1) activation models, (2) output shapes per layer, (3) dropout layers, and (4) early stopping values. The test results show that RELU activation is better than Sigmoid. Rescaling (128x128) is better for scala (64x64) and (256x256) which affects the output shape model of each layer. In this learning stage, the use of dropouts in the CNN architecture achieves robust accuracy than the architecture that ignores dropouts. The use of 15 early stoppings is better than other values compared. 20 images of pneumonia and 20 images of covid have been tested using the proposed method and achieved 87.50% accuracy, 80.00% precision, 100% recall, and 99.89% F1-Score. Our method is superior to the the comparison method in terms of accuracy, precision, recall, and f1-score, which achieves 85%, 70%, 100%, and 82.35%, respectively
UML Transformation to Java-based Software Product Lines
Full text linkSoftware product line engineering (SPLE) is an emerging approach that enables variability management in software development. SPLE offers tremendous benefits, but lack of tool support becomes a barrier in the adoption of SPLE. Variability modules for Java (VMJ) is an implementation approach that is defined based on the variability modules (VM) concept to support SPLE. VMJ combines Java modules system and design patterns that are commonly used by software developers. VMJ is accompanied by a UML profile, called UML-VM profile, which extends UML notation to model variability in the UML diagram. UML-VM diagram is used to model the problem domain, and VMJ is used in the domain implementation. In this research, we design a model transformation from Unified Modeling Language (UML) diagram into VMJ. The transformation rules are defined based on the UML-VM profile and implemented in the Eclipse Acceleo model to text transformation. As a result, a UML diagram can be transformed automatically into Java-based software product lines. The transformation tool is evaluated using a case study by comparing the generated code and the actual implementation
Sentiment Analysis of COVID-19 Vaccines in Indonesia on Twitter Using Pre-Trained and Self-Training Word Embeddings
Full text linkSentiment analysis regarding the COVID-19 vaccine can be obtained from social media because users usually express their opinions through social media. One of the social media that is most often used by Indonesian people to express their opinion is Twitter. The method used in this research is Bidirectional LSTM which will be combined with word embedding. In this study, fastText and GloVe were tested as word embedding. We created 8 test scenarios to inspect performance of the word embeddings, using both pre-trained and self-trained word embedding vectors. Dataset gathered from Twitter was prepared as stemmed dataset and unstemmed dataset. The highest accuracy from GloVe scenario group was generated by model which used self-trained GloVe and trained on unstemmed dataset. The accuracy reached 92.5%. On the other hand, the highest accuracy from fastText scenario group generated by model which used self-trained fastText and trained on stemmed dataset. The accuracy reached 92.3%. In other scenarios that used pre-trained embedding vector, the accuracy was quite lower than scenarios that used self-trained embedding vector, because the pre-trained embedding data was trained using the Wikipedia corpus which contains standard and well-structured language while the dataset used in this study came from Twitter which contains non-standard sentences. Even though the dataset was processed using stemming and slang words dictionary, the pre-trained embedding still can not recognize several words from our dataset
SGCF: Inductive Movie Recommendation System with Strongly Connected Neighborhood Sampling
Full text linkUser and item embeddings are key resources for the development of recommender systems. Recent works has exploited connectivity between users and items in graphs to incorporate the preferences of local neighborhoods into embeddings. Information inferred from graph connections is very useful, especially when interaction between user and item is sparse. In this paper, we propose graphSAGE Collaborative Filtering (SGCF), an inductive graph-based recommendation system with local sampling weight. We conducted an experiment to investigate recommendation performance for SGCF by comparing its performance with baseline and several SGCF variants in Movielens dataset, which are commonly used as recommendation system benchmark data. Our experiment shows that weighted SGCF perform 0.5% higher than benchmark in NDCG@5 and NDCG@10, and 0.8% in NDCG@100. Weighted SGCF perform 0.79% higher than benchmark in recall@5, 0.4% increase for recall@10 and 1.85% increase for recall@100. All the improvements are statistically significant with p-value < 0.05
Gender Prediction of Indonesian Twitter Users Using Tweet and Profile Features
Full text linkThe increasing use of social media generates huge amounts of data which in turn triggers research into social media analytics. Social media contents can be analyzed to explore public opinion on an issue or provide the insights reflecting proxy indicators towards real-world events. Understanding the demographics of social media users can increase the potential for applications of sentiment analysis, topic modeling, and other analytical tasks. To map demographics, we need to know the latent attributes of users, such as age, gender, occupation and location of residence. Since this attribute is not directly available, we need to do some inference from the social media data. This study aims to predict the gender attribute given a Twitter user account. We conducted experiments with several supervised classifiers with feature extraction, including the use of word embedding representations. The results of this study indicate that the combination of features extracted from Tweet contents and user profile structured data can predict the gender of Twitter users in Indonesia with accuracy above 80%
Wavelet Transformation and Spectral Subtraction Method in Performing Automated Rindik Song Transcription
Full text linkRindik is Balinese traditional music consisting of bamboo rods arranged horizontally and played by hitting the rods with a mallet-like tool called "panggul". In this study, the transcription of Rindik's music songs was carried out automatically using the Wavelet transformation method and spectral subtraction. Spectral subtraction method is used with iterative estimation and separation approaches. While the Wavelet transformation method is used by matching the segment Wavelet results with the Wavelet result references in the dataset. The results of the transcription were also synthesized again using the concatenative synthesis method. The data used is the hit of 1 Rindik rod and a combination of 2 Rindik rods that are hit simultaneously, and for testing the system, 4 Rindik songs are used. Each data was recorded 3 times. Several parameters are used for the Wavelet transformation method and spectral subtraction, which are the length of the frame for the Wavelet transformation method and the tolerance interval for frequency difference in spectral subtraction method. The test is done by measuring the accuracy of the transcription from the system within all Rindik song data. As a result, the Wavelet transformation method produces an average accuracy of 83.42% and the spectral subtraction method produces an average accuracy of 78.51% in transcription of Rindik songs
Epidemic Data Analysis of Three Variants of COVID-19 Spread in Indonesia
Full text linkThree variants of COVID-19 had been found in Indonesia. A control strategy may rely on the transmission rate of the variant. This study aims to investigate how the variants spread in Indonesia by computing a basic and effective reproduction number on the national and province scale. The basic reproduction number shows the indicator of initial transmission rate of alpha variant computed by an exponential growth rate model. The effective reproduction number describes the dynamic of the transmission rate estimated based on a Bayesian approach. This study revealed that each variant shows different characteristics. The alpha variant of COVID-19 in Indonesia was mainly initiated from big cities, then it spread to all provinces quickly because the control strategies were not established well at the beginning. A rapid increase of the effective reproduction number about July 2021 showed a novel delta variant, but it could be managed quite well by a large number of testing and stronger restrictions. Before the end of 2021, a novel variant omicron was also shown by the steeper change of the effective reproduction number. Thus, the variant spread rate can be estimated by how steep the effective reproduction number change is