Jurnal Ilmu Komputer dan Informasi
Not a member yet
247 research outputs found
Sort by
A Systematic Literature Review on SOTA Machine learning-supported Computer Vision Approaches to Image Enhancement
Full text linkImage enhancement as a problem-oriented process of optimizing visual appearances to provide easier-toprocess input to automated image processing techniques is an area that will consistently be a companion to computer vision despite advances in image acquisition and its relevance continues to grow. For our systematic literature review, we consider the major peer-reviewed journals and conference papers on the state of the art in machine learning-based computer vision approaches for image enhancement. We describe the image enhancement methods relevant to our work and introduce the machine learning models used. We then provide a comprehensive overview of the different application areas and formulate research gaps for future scientific work on machine learning based computer vision approaches for image enhancement based on our result
Bimodal Keystroke Dynamics-Based Authentication for Mobile Application Using Anagram
Full text linkCurrently, most of the smartphones recognize uses based on static biometrics, such as face and fingerprint. However, those traits were vulnerable against spoofing attack. For overcoming this problem, dynamic biometrics like the keystroke and gaze are introduced since it is more resistant against spoofing attack. This research focuses on keystroke dynamics for strengthening the user recognition system against spoofing attacks. For recognizing a user, the user keystrokes feature used in the login process is compared with keystroke features stored in the keystroke features database. For evaluating the accuracy of the proposed system, words generated based on the Indonesian anagram are used. Furthermore, for conducting the experiment, 34 participants were asked to type a set of words using the smartphone keyboard. Then, each user’s keystroke is recorded. The keystroke dynamic feature consists of latency and digraph which are extracted from the record. According to the experiment result, the error of the proposed method is decreased by 23.075% of EER with FAR and FRR are decreased by 16.381% and 10.41% respectively, compared with Kim’s method. It means that the proposed method is successful increase the biometrics performance by reducing the error rate
Improving Recognition of SIBI Gesture by Combining Skeleton and Hand Shape Features
Full text linkSIBI (Sign System for Indonesian Language) is an official sign language system used in school for hearing impairment students in Indonesia. This work uses the skeleton and hand shape features to classify SIBI gestures. In order to improve the performance of the gesture classification system, we tried to fuse the features in several different ways. The accuracy results achieved by the feature fusion methods are, in descending order of accuracy: 88.016%, when using sequence-feature-vector concatenation, 85.448% when using Conneau feature vector concatenation, 83.723% when using feature-vector concatenation, and 49.618% when using simple feature concatenation. The sequence-feature-vector concatenation techniques yield noticeably better results than those achieved using single features (82.849% with skeleton feature only, 55.530% for the hand shape feature only). The experiment results show that the combined features of the whole gesture sequence can better distinguish one gesture from another in SIBI than the combined features of each gesture frame. In addition to finding the best feature combination technique, this study also found the most suitable Recurrent Neural Network (RNN) model for recognizing SIBI. The models tested are 1-layer, 2-layer LSTM, and GRU. The experimental results show that the 2-layer bidirectional LSTM has the best performance
Predicting Analysis of User’s Interest from Web Log Data in e-Commerce using Classification Algorithms
Full text linkThe accelerated development of e-commerce has been a concern for business people. Business people should be able to gain customer interest in a variety of ways so that their companies can compete with others. Analyzing click-flow data will help organizations or firms assess customer loyalty, provide advertising privileges, and develop marketing strategies through user interests. By understanding consumer preferences, clickstream data analysis may be used to determine who is participating, assist companies in evaluating customer contentment, boost productivity, and design marketing strategies. This research was performed by defining experimental user interests using Dynamic Mining and Page Interest Estimation methods. The findings of this analysis, using three algorithms at the pattern discovery page, demonstrated that the Decision Tree method excelled in both methods. It indicated that the operational performance of the Decision Tree performed well in the assessment of user interests with two different approaches. The findings of this experiment can be used as a proposal for researching the field of web usage mining, collaborating with other approaches to achieve higher accuracy values
Myers-Briggs Type Indicator Personality Model Classification in English Text using Convolutional Neural Network Method
Full text linkMyers-Briggs Type Indicator (MBTI) is a personality model developed by Katharine Cooks Briggs and Isabel Briggs Myers in 1940. It displays a combination of preferences from four domains. Generally, test takers need to answer about 50 to 70 questions, and it is relatively expensive to know MBTI personality. The researcher developed a personality classification system using the Convolutional Neural Network (CNN) method and GloVe (Global Vectors for Word Representation) word embedding to solve this problem. The dataset used in this research consists of 8,675 data from the Kaggle site. The steps in this research are downloading the dataset from Kaggle, text preprocessing, GloVe weighting, classification using the CNN method, and evaluation using accuracy from the Confusion Matrix. Based on the tests carried out, using GloVe weighting can improve the model accuracy rather than random weighting. The best GloVe word dimensions depend on the metrics used to measure the model performance and the data of the classes contained in the dataset. From the CNN hyperparameter tuning test, the Adamax optimizer performs better and produces higher accuracy than the Adam optimizer. In addition, the CNN hyperparameter tuning increased model accuracy more significantly compared with the best GloVe word embedding dimensions
Towards Erlang-based ABS Microservices Framework for Software Product Line Development
Full text linkThe current widely used software system can be categorised as a large or very large decentralised control system with various requirements and continuous interchangeable elements. This characteristic leads to a need to control the variability to manage such systems. Software Product Line Engineering (SPLE) is one of the approaches that can manage the variability by developing sets of products. However, there is a need for support tools for development with software product line engineering. One language that supports the SPLE process is Abstract Behavioral Specification (ABS). Some SPLE research has used ABS to create frameworks that support the SPLE process. ABS Microservices is one research that utilises ABS to create a web framework that supports the SPLE process. This framework uses ABS to generate Java-based applications. The research interest in the web application is driven by the fact that it is one of the software types widely used by organisations and serves as the primary support of their business. Microservices are highly interoperable, thus enabling researchers to integrate different technology from other research. However, there is a need for renewal to the ABS Microservices framework. There is a need for more variants of SPLE-enabled frameworks that use more programming language as a specific programming language has its strength and weakness. Deprecation of the Java backend of the ABS opens a new exploration of another web framework that uses other ABS backend languages. We present the ABS microservices web framework based on Erlang OTP. We choose Erlang because it promises more efficient resource usage and the Erlang backend is one of the ABS backends with the most available features. This research aims to create an entry point for ABS Microservices to support more language. This research shows that the Erlang variant of ABS Microservices has less resource usage than the Java variant. Hence, this promises more options to develop product lines using ABS Microservices
Analysis of Livestock Meat Production in Indonesia Using Fuzzy C-Means Clustering
Full text linkThe production of livestock in Indonesia is one type of food that the public can consume. Indonesia is still importing meat for food for its people. This study aims to classify provinces in Indonesia with high livestock meat production and low livestock meat production so that the government can maximize areas with high livestock meat production and can seek to increase livestock meat production in areas with low production. Clustering is needed to identify groups of livestock meat-producing provinces with high and low production. The data is grouped into 2 clusters using FCM with a silhouette index value of 0.95664, the first cluster with the highest meat production total in three provinces (West Java, Central Java, and East Java) and the second cluster with the lowest meat production total 31 provinces. West Java, Central Java, and East Java mostly work as livestock breeders due to the availability of sufficient land
Reducing Adversarial Vulnerability through Adaptive Training Batch Size
Full text linkNeural networks possess an ability to generalize well to data distribution, to an extent that they are capable of fitting to a randomly labeled data. But they are also known to be extremely sensitive to adversarial examples. Batch Normalization (BatchNorm), very commonly part of deep learning architecture, has been found to increase adversarial vulnerability. Fixup Initialization (Fixup Init) has been shown as an alternative to BatchNorm, which can considerably strengthen the networks against adversarial examples. This robustness can be improved further by employing smaller batch size in training. The latter, however, comes with a tradeoff in the form of a significant increase of training time (up to ten times longer when reducing batch size from the default 128 to 8 for ResNet-56). In this paper, we propose a workaround to this problem by starting the training with a small batch size and gradually increase it to larger ones during training. We empirically show that our proposal can still improve adversarial robustness (up to 5.73\%) of ResNet-56 with Fixup Init and default batch size of 128. At the same time, our proposal keeps the training time considerably shorter (only 4 times longer, instead of 10 times)
Brain Tumors Detection By Using Convolutional Neural Networks and Selection of Thresholds By Histogram Selection
Full text linkBrain tumors in medical images have a high diversity in terms of shape and size. Some of the data found a form between the tumor tissue and normal tissue, whereas knowing the tumor’s profile and characteristics becomes a crucial part of searching. By using machine learning capabilities, where machines are given several variables and provide decisions to a certain degree, they have broadly given decisions that support subject matter in making decisions. This study applies the threshold selection method using histogram selection on CT scan data, while the appropriate threshold selection method selects the tumor position accordingly. Furthermore, the Convolutional Neural Network (CNN) is used to classify whether the selected image is a tumor or not. Using CT scan data and calculated experiments, this algorithm can finally be approved and given a brain classification with an accuracy of 75.42 percent
The Stock Exchange Prediction using Machine Learning Techniques: A Comprehensive and Systematic Literature Review
Full text linkThis literature review identifies and analyzes research topic trends, types of data sets, learning algorithm, methods improvements, and frameworks used in stock exchange prediction. A total of 81 studies were investigated, which were published regarding stock predictions in the period January 2015 to June 2020 which took into account the inclusion and exclusion criteria. The literature review methodology is carried out in three major phases: review planning, implementation, and report preparation, in nine steps from defining systematic review requirements to presentation of results. Estimation or regression, clustering, association, classification, and preprocessing analysis of data sets are the five main focuses revealed in the main study of stock prediction research. The classification method gets a share of 35.80% from related studies, the estimation method is 56.79%, data analytics is 4.94%, the rest is clustering and association is 1.23%. Furthermore, the use of the technical indicator data set is 74.07%, the rest are combinations of datasets. To develop a stock prediction model 48 different methods have been applied, 9 of the most widely applied methods were identified. The best method in terms of accuracy and also small error rate such as SVM, DNN, CNN, RNN, LSTM, bagging ensembles such as RF, boosting ensembles such as XGBoost, ensemble majority vote and the meta-learner approach is ensemble Stacking. Several techniques are proposed to improve prediction accuracy by combining several methods, using boosting algorithms, adding feature selection and using parameter and hyper-parameter optimization