IJCCS (Indonesian Journal of Computing and Cybernetics Systems)
Not a member yet
480 research outputs found
Sort by
Automatic Detection of Helmets on Motorcyclists Using Faster - RCNN
Get PDFMotorcycles have been a popular choice for a go-to daily means of transportation due to its lower price, making it affordable for high to low-class citizens. Helmets are required for every motorcycle owner so that the rider’s head is protected from accidents. However, not many people follow the rules and tend to not wear helmets and plenty of them underestimate the usage of helmets. For this, it is necessary to implement a system that can detect which rider wears the helmet or not by applying deep learning techniques. This paper aims to implement one of the deep learning techniques, which is Faster R – CNN to detect the helmets and the motorcyclists. After training 400 images using different learning rates, the mean average precision (mAP) achieved the highest with 87% using the learning rate of 0.000
Neural Network Pruning in Unsupervised Aspect Detection based on Aspect Embedding
Get PDF Aspect detection systems for online reviews, especially based on unsupervised models, are considered better strategically to process online reviews, generally a very large collection of unstructured data. Aspect embedding-based deep learning models are designed for this problem however they still rely on redundant word embedding and they are sensitive to initialization which may have a significant impact on model performance. In this research, a pruning approach is used to reduce the redundancy of deep learning model connections and is expected to produce a model with similar or better performance. This research includes several experiments and comparisons of the results of pruning the model network weights based on the general neural network pruning strategy and the lottery ticket hypothesis. The result of this research is that pruning of the unsupervised aspect detection model, in general, can produce smaller submodels with similar performance even with a significant amount of weights pruned. Our sparse model with 80% of its total weight pruned has a similar performance to the original model. Our current pruning implementation, however, has not been able to produce sparse models with better performance
Naive Bayes Method and C4.5 in Classification of Birth Data
Get PDFData on the birth and productive age of a mother to get pregnant in Lampung is still high. to find out the comparison of the productive age of pregnant women and whether they have met the minimum and maximum requirements for a mother to become pregnant, and the criteria for babies born. Where the results of data processing will be used as a source of data for counseling mothers, especially for residents of Banjar Kertahayu village. The data processing requires a special method so that the results become a benchmark for a decision later, such as Data Mining. The method used for data processing used is Naive Bayes and C4.5 Algorithm. The data used is birth data in 2017-2021, the source of data from the Banjar Village Midwife-Central Lampung Regency. Research Results Method C 4.5 Middle age has a dominant age category value of 0.3324138. where the highest value is in 2017, and accuracy is 100 percent from the 2017-2021 data. The baby weight criterion using the Naïve Bayes Class Method has a dominant Middle-aged category value of 0.09675, the highest value in 2017, The results of accuracy for 5 years have accuracy of 92.84% based on 2017-2021 birth dat
Aspect-Based Sentiment Analysis of KAI Access Reviews Using NBC and SVM
Get PDFThe existence of KAI Access from PT. KAI prove their sincerity in serving consumers in this modern era. However, many negative reviews found in Google Play Store. There has been research on the review, but the analysis stage still at document level so the aspect related to the application is not known clearly and structured. So it is necessary to do an aspect-based sentiment analysis to extract the aspects and the sentiment. This study aims to do an aspect-based sentiment analysis on user reviews of KAI Access using Naive Bayes Classifier (NBC) and Support Vector Machine (SVM), with 3 scenarios. Scenario 1 uses NBC with Multinomial Naive Bayes, scenario 2 uses SVM with default Sklearn library parameter, and scenario 3, uses SVM with hyperparameter tunning, while the data scrapped from Google Play Store. The results show the majority of user sentiment is negative for each aspect, with most discussed errors aspect shows the high system errors. The test results gives the best model from scenario 3 with an average accuracy 91.63%, f1-score 75.55%, precision 77.60%, and recall 74.47%
Ontology-based Complementary Breastfeeding Search Model
Get PDFChildren's nutritional requirements differ from those of adults. The health ministry's Indonesian data shows that in 2017, there were 17.8% of malnourished children under five years old (toddlers), one of which was related to complementary breastfeeding problems. Complementary breastfeeding is given to babies starting at 6–24 months of age. This research aims to build a complementary breastfeeding search model and be able to present it as a treatment for malnourished babies. A search model is built to understand natural language input given by a user. Also, it can do reasoning by applying a set of rules to obtain implicit knowledge about the complementary breastfeeding menu recommended for babies. The methods used in this research are data collection, designing a search model, building an ontology model, building SWRL, natural language processing, and usability testing by users and nutritionists. This research succeeded in building an ontology-based complementary breastfeeding search model in the form of a semantic web. The testing result shows that the web can provide an alternative complementary breastfeeding menu according to the baby’s nutritional needs and has a high usability capability of 4.01 on a scale of 1 to 5
Face Expression Classification in Children Using CNN
Get PDFOne of the turbulent emotions can be recognized from facial expressions. When compared with adults, children's facial expressions are more expressive for positive emotions and ambiguous for negative emotions so that they are much more difficult to recognize. Ambiguous in terms of negative emotions, for example, when children are angry, sometimes they show an expressionless face, making it difficult to know what emotions the child is experiencing. Therefore, it is proposed research using Convolutional Neural Network with ResNet-50 architecture. According to [1] CNN Resnet-50 is superior to other facial recognition methods, specifically in the classification of facial expressions. CNN ResNet-50 generates a model during the training process, and the model will be used during the testing process. The dataset used is Children's Spontaneous facial Expressions (LIRIS-CSE) data proposed by [2]. CNN ResNet-50 can identify children's expressions well, including expressions of anger, disgust, fear, happy, sad and surprise. The results showed a very significant increase in accuracy, namely in testing data testing reached 99.89%
Internet of Things (IoT) Arduino-Based Classroom Monitoring Utilizes Temperature Sensors And CO2 Sensors
Get PDFComfort room temperature is determined by indoor air quality, such as temperature and CO2 gas. This study aims to determine the comfort of a class by reviewing the number of students, CO2 gas, and temperature in an Arduino-based classroom using an automatic IoT system with a Completely Randomized Design (CRD) method. Research proves that there is a significant effect between the number of students on the concentration of CO2, but it does not directly affect the air temperature in the room. The lecture hall is still relatively safe but not ideal and requires a temperature reduction of -7oC
Comparison of SVM and LIWC for Sentiment Analysis of SARA
Get PDFSARA is a sensitive issue based on sentiments about self-identity regarding ancestry, religion, nationality or ethnicity. The impact of the issue of SARA is conflict between groups that leads to hatred and division. SARA issues are widely spread through social media, especially Twitter. To overcome the problem of SARA, it is necessary to develop an effective method to filter negative SARA. This study aims to analyze Indonesian-language tweets and determine whether the tweet contains positive or negative SARA or does not contain SARA (neutral). Machine learning (i.e., SVM) and lexicon-based method (i.e., LIWC) were compared based on 450 tweet data to determine the best approach for each sentiment (positive, negative, and neutral). The best evaluation results are shown in the negative SARA classification using SVM with λ = 3 and γ = 0.1, where Precision = 0.9, Recall = 0.6, and F1-Score = 0.72. The best results from the positive SARA classification were shown in the LIWC method, where Precision = 0.6, Recall = 0.8, and F1-Score = 0.69. The best evaluation results for neutral classification are shown in SVM with λ = 3 and γ = 0.1, with Precision = 0.52, Recall = 0.87, and F1-Score = 0.65
Implementation of Factor Analysis and BiClustering in Classifying Multidimensional Under-Five Poverty in East Nusa Tenggara
Get PDFUnder-five poverty is a condition where the needs of toodlers are not met, resulting in undernourished children and unable to reach their full potential in the social sphere. East Nusa Tenggara is a province that still faces the biggest nutritional problems in Indonesia in 2019. This study aims to explain the variables that form toodlers multidimensional poverty in East Nusa Tenggara (ENT), form the Multidimensional Under-Five Poverty Index (MUPI), and compare the results of index formed with the results of bicluster. Data source used in this study is SUSENAS KOR 2019. The analytical method used is a factor and bicluster analysis. The results shows that 11 multidimensional poverty indicators form three dimensions, namely the Adequate Food and Beverage Facility Factor, Health Protection Factor, and Housing and Nutrition Factor, which is used to form the index. Based on regional grouping, there are five areas with low MUPI scores, fourteen areas with medium MUPI scores, and three areas with high MUPI scores. However, biclustering results show that there are two areas with low poverty category, thirteen regions with moderate poverty category, and seven regions with high poverty category. The result of the comparison of MUPI grouping with the biclustering method obtained different results based on the composition of the resulting area
Application of the Weighted Product Method in a Decision Support System to Determine Children's Multiple Intelligence
Get PDFIntelligence cannot solely be measured in terms of intellectual intelligence. There are various types of intelligence in children, which cause teachers and parents require time to determine the type of intelligence of children. Quick and easy decision-making can be achieved using a decision support system. One method that can be adopted in the decision-making process of a decision support system is the weighted product method. This study aims to measure the level of accuracy of the weighted product method in determining the type of multiple intelligence of children. The decision support system determines the type of intelligence of children in early childhood (ages 4-6 years) using Garner's [1] eight types of multiple intelligences as decision-making criteria. The data was collected using interviews and questionnaires to the teachers of Mutiara State Kindergarten. The study found that a decision support system using the weighted product method can determine the type of children's multiple intelligences with an accuracy rate of 96%. Based on the result of analysis and calculation using the weighted product method from test questionnaire data of 55 children, compared to the results of identification by the teacher, it was found that the compatibility of 53 children