Search

In synchronous online lectures, the lecturers often provide the lecture material directly through video conference technology. On the other hand, there are many students who do not pay attention to the lecturers when they are participating in online lectures. As a consequence, in this research, an application was developed to assist lecturers in gathering data regarding the degree to which students who participate in online lectures pay attention to the presented information. The application employed a convolutional neural network (CNN) model to recognize each student's facial expressions and place them into one of two classes: either engaged or disengaged. The captured student facial image was preprocessed to facilitate the classification process. The preprocessing stage consisted of image conversion to gray scale, face detection using the Haar-Cascade Classifier model, and a median filter to reduce noise. In the process of designing a CNN model, three different hyperparameter tuning scenarios were implemented. These tuning scenarios aimed to obtain the best possible CNN model by determining which CNN model hyperparameters were the most optimal. The results of the experiments indicate that the CNN model from the second scenario has the highest level of accuracy in terms of recognizing facial expressions, coming in at 86%. The results of this research have been tested to measure the level of student participation in online lectures. The trial results show that the proposed application can help lecturers evaluate student engagement during online lectures.

Several recent studies in dentistry and maxillofacial imaging have concentrated on the Mandibular Canal (MC) segmentation in digital dental radiographs. In this research domain, deep learning approaches have demonstrated promising outcomes. This systematic literature review (SLR) aims to comprehensively analyze and synthesize the recent advancements in applying deep learning techniques for MC segmentation in digital dental radiographs. This review encompasses studies published between 2018 and 2023, sourced from reputable databases, including PubMed, ScienceDirect, IEEE Xplore, Google Scholar, and SpringerLink. This study identified 30 primary research papers focusing on MC segmentation in digital dental radiographs. The review categorizes papers into two groups based on the digital dental radiograph type. The first group uses a 2D digital dental radiograph from a panoramic radiograph and 2D Cone Beam Computed Tomography (CBCT) scans. The second group uses 3D datasets from volumetric data from CBCT scans. The synthesized knowledge from this review is intended to guide researchers, dentists, and oral surgeons in leveraging deep learning advancements in MC segmentation for oral and maxillofacial surgery. Prior studies have faced challenges, including limited public datasets, variations in MC anatomy, time consumption and observer variability in MC annotation, complexities in deep learning models, and lack of practical implementation. To overcome these challenges, it is suggested that more public datasets be collaboratively created and shared within the research community, focusing on handling anatomy variability, improving digital dental radiograph quality, streamlining annotation processes through automated tools, and simplifying deep learning models for practical implementation.

Background: The learning assessment of elementary schools has recently incorporated Bloom's Taxonomy, a structure in education that categorizes different levels of cognitive learning and thinking skills, as a fundamental framework. This assessment now includes High Order Thinking Skill (HOTS) questions, with a specific focus on Indonesian topics. The implementation of this system has been observed to require teachers to manually categorize or classify questions, and this process typically requires more time and resources. To address the associated difficulty, automated categorization and classification are required to streamline the process. However, despite various research efforts in questions classification, there is still room for improvement in terms of performance, particularly in precision and accuracy. Numerous investigations have explored the use of Deep Learning Natural Language Processing models such as BERT for classification, and IndoBERT is one such pre-trained model for text analysis. Objective: This research aims to build classification system that is capable of classifying Indonesian exam questions in multiple-choice form based on Bloom's Taxonomy using IndoBERT pre-trained model. Methods: The methodology used includes hyperparameter fine-tuning, which was carried out to identify the optimal model performance. This performance was subsequently evaluated based on accuracy, F1 Score, Precision, Recall, and the time required for the training and validation of the model. Results: The proposed Fine Tuned IndoBERT Model showed that the accuracy rate was 97%, 97% F1 Score, 97% Recall, and 98% Precision with an average training time per epoch of 1.55 seconds and an average validation time per epoch of 0.38 seconds. Conclusion: Fine Tuned IndoBERT model was observed to have a relatively high classification performance, and based on this observation, the system was considered capable of classifying Indonesian exam questions at the elementary school level. Keywords: IndoBERT, Fine Tuning, Indonesian Exam Question, Model Classifier, Natural Language Processing, Bloom’s Taxonomy

Obat merupakan bahan kimia yang dapat merepresentasikan tubuh secara fisiologi dan psikologi ketika dikonsumsi. Obat sebagai alat bantu untuk menyembuhkan dari berbagai macam penyakit. Dengan berkembangnya zaman dan bertambahnya wawasan, menyebabkan bertambah juga jenis obat-obatan yang memiliki banyak manfaat dan kegunaanya. Penelitian ini bertujuan untuk mendeteksi nama obat dalam resep dokter menggunakan Convolutional Neural Network (CNN) dengan transfer learning. Metode transfer learning merupakan metode yang popular dalam mengklasifikasi gambar digital yang berguna untuk mempercepat proses klasifikasi. Penelitian ini membandingkan lima artistektur transfer learning yaitu VGG16, Resnet, Xception, LeNet, dan GoogleNet. Penelitian ini juga menggunakan grayscaling, resizing, dan median filter pada tahap preprocessing. Preprocessing digunakan untuk meningkatkan kualitas citra pada citra resep obat dan menghilangkan noise pada citra. ResNet-50 merupakan arsitektur terbaik untuk mengklasifikasi nama obat. Pada percobaan menggunakan ResNet-50, mendapatkan F1 score tertinggi yaitu sebesar 97,56% dan waktu training rata-rata 0,25 detik setiap epoch. Dapat disimpulkan Resnet merupakan arsitektur terbaik untuk mengklasifikasikan nama obat dalam citra resep dokter serta dapat mendeteksi nama obat secara akurat.

Nasabah adalah salah satu aset paling berharga dari sebuah bisnis perbankan. Mereka adalah ujung tombak pengguna produk yang nantinya memberikan keuntungan bagi bank, terutama pada produk kartu kredit. Penelitian ini bertujuan untuk mengetahui nasabah mana sajakah yang berpotensi untuk meninggalkan layanan kartu kredit dari sebuah bank. Pada penelitian sebelumnya belum ada yang melakukan analisis perbandingan algoritma machine learning dengan berbagai macam tahapan preprocessing untuk memprediksi potensi hilangnya nasabah bank. Penelitian ini melakukan analisis perbandingan algoritma machine learning dengan kombinasi tahapan preprocessing untuk memprediksi potensi hilangnya nasabah bank. Analisis ini penting untuk pemilihan algoritma yang paling cocok untuk prediksi potensi hilangnya nasabah bank. Pada tahapan preprocessing diterapkan dimensionality reduction dan feature selection menggunakan metode Variance threshold dan Correlation coefficient. Metode klasifikasi yang digunakan adalah Logistic regression (LR), Decision tree (DT), dan Naïve Bayes (NB). Hasil tertinggi dari ketiga metode tersebut adalah Decision tree yang mampu memiliki nilai F1 Score sebesar 96% dan nilai akurasi mencapai 93%. Logistic regression dan Naïve Bayes berada pada urutan kedua dan ketiga setelah decision tree. Tahapan data preprocessing tidak memberikan pengaruh yang signifikan pada nilai F1 Score dan akurasi.

Background: Melanoma is a skin cancer that starts when the melanocytes that produce the skin color pigment start to grow out of control and form a cancer. Detecting melanoma early before it spreads to the lymph nodes and other parts of the body is very important because it makes a big difference to the patient's 5-year life expectancy. Screening is the process of conducting a skin examination to suspect a mole is melanoma using dermoscopic or macroscopic images. However, manual screening takes a long time. Therefore, automatic melanoma detection is needed to speed up the melanoma detection process. The previous studies still have weakness because it has low precision or recall, which means the model cannot predict melanoma accurately. The distribution of melanoma and moles datasets is imbalanced where the number of melanomas is less than moles. In addition, in previous study, comparisons of several CNN transfer learning architectures have not been carried out on dermoscopic and macroscopic images. Objective: This study aims to detect melanoma using the Convolutional Neural Network (CNN) with transfer learning on dermoscopic and macroscopic melanoma images. CNN with Transfer learning is a popular method for classifying digital images with high accuracy. Methods: This study compares four CNN with transfer learning architectures, namely MobileNet, Xception, VGG16, and ResNet50 on dermoscopic and macroscopic image. This research also uses black-hat filtering and inpainting at the preprocessing stage to remove hair from the skin image. Results: MobileNet is the best model for classifying melanomas or moles in this experiment which has 83.86% of F1 score and 11 second of training time per epoch. Conclusion: MobileNet and Xception have high average F1 scores of 84.42% and 80.00%, so they can detect melanoma accurately even though the number of melanoma datasets is less than moles. Therefore, it can be concluded that MobileNet and Xception are suitable models for classifying melanomas and moles. However, MobileNet has the fastest training time per epoch which is 11 seconds. In the future, oversampling method can be implemented to balance the number of datasets to improve the performance of the classification model.

Terdapat orang yang tidak mampu berkomunikasi secara verbal yang menyebabkan kesulitan dalam berkomunikasi. Orang tersebut mengalami gangguan seperti tuli atau bisu. Mereka hanya dapat berkomunikasi melalui bahasa isyarat salah satunya adalah Sistem Isyarat Bahasa Indonesia (SIBI). Pengenalan Bahasa Isyarat adalah permasalahan klasifikasi yang kompleks untuk dipecahkan. Setiap bahasa isyarat memiliki sintaks dan tata bahasanya sendiri. Computer vision adalah sebuah Teknik yang digunakan komputer untuk melakukan klasifikasi citra. Computer vision membantu pengenalan citra SIBI secara otomatis sehingga memudahkan orang normal berkomunikasi dengan orang tuli atau bisu. Pada penelitian sebelumnya belum ada yang melakukan perbandingan algoritma klasifikasi machine learning dan deep learning untuk pengenalan SIBI. Perbandingan penting dilakukan untuk melihat efektifitas tiap algoritma klasifikasi dalam hal performa klasifikasi dan waktu komputasi. Algoritma klasifikasi machine learning memiliki waktu komputasi lebih rendah sedangkan Deep learning memiliki performa klasifikasi lebih tinggi. Penelitian ini menganalisis time to performance dari algoritma machine learning dan deep learning dalam melakukan klasifikasi citra SIBI huruf A hingga Z. K-Nearest Neighbors (KNN), Support Vector Machine (SVM), dan Convolutional neural network (CNN) dengan transfer learning adalah tiga algorimta klasifikasi populer yang dibandingkan dalam penelitian ini. Arsitektur transfer learning yang digunakan adalah Xception, ResNet50, VGG15, dan MobileNetV2. Dari hasil penelitian yang dilakukan menggunakan 5 cross validation, CNN dengan arsitektur Xception memiliki nilai F1 Score tertinggi yaitu 99,57% dengan waktu training rata-rata 1.387 detik. Sedangkan KNN dengan nilai K = 1 memiliki waktu training tercepat yaitu 0,03 detik dan memiliki nilai F1 Score 86,95%. Abstract The person who has a disorder such as deaf or dumb are unable to communicate verbally, which causes difficulties in communicating. They can only communicate through sign language, one of which is the Indonesian Language Sign System or Sistem Isyarat Bahasa Indonesia (SIBI). Sign Language Recognition is a complex classification problem to solve. Each sign language has its syntax and grammar. Computer vision is a technique used by computers to classify images. Computer vision helps automatically recognize SIBI images, making it easier for normal people to communicate with deaf or mute people. In previous studies, no one has compared machine learning and deep learning classification algorithms for the classification of SIBI. Therefore, a meaningful comparison is made to see each classification algorithm's effectiveness in classification performance and computation time. Machine learning classification algorithms have lower computation time, while Deep learning has higher classification performance. This study analyzes the time to performance of machine learning and deep learning algorithms in classifying SIBI images of letters A to Z. K-Nearest Neighbors (KNN), Support Vector Machine (SVM), and Convolutional neural network (CNN) with transfer learning are three popular classification algorithms compared in this study. The transfer learning architectures used are Xception, ResNet50, VGG15, and MobileNetV2. The results of research conducted using 5 cross-validation, CNN with the Xception architecture has highest F1 Score of 99.57%, with an average training time of 1.387 seconds. KNN, with a value of K = 1, has the fastest training time of 0.03 seconds and an F1 Score of 86.95%.

Buah tomat merupakan salah satu buah yang sering dijumpai di Indonesia. Permintaan tomat juga terus meningkat dari tahun ke tahun di Indonesia. Keterbatasan lahan menjadi kendala dalam memenuhi permintaan tersebut. Oleh karena itu, tomat dapat dikembangkan melalui media kultur jaringan untuk menghasilkan buah yang sehat dan cepat. Salah satu faktor yang menentukan keberhasilan dalam budidaya tomat melalui media kultur jaringan adalah intensitas cahaya yang digunakan. Metode multiple linear regression diimplementasikan untuk dapat memperkirakan intensitas cahaya yang tepat. Fitur yang akan menjadi prediktor dalam penelitian ini adalah jenis media tanam dan panjang kecambah yang diharapkan. Berdasarkan evaluasi yang telah dilakukan, nilai mean absolute error (MAE) yang dihasilkan yaitu 0,87 dengan rasio data training dan testing adalah 90:10. Hal ini menandakan bahwa sistem cukup akurat dalam melakukan prediksi.

Penelitian tentang otomatisasi pembangkitan pertanyaan terus berkembang. Berbagai metode telah coba diterapkan pada berbagai bahasa. Setiap bahasa memiliki karakteristik yang berbeda beda. Oleh karena itu, metode yang digunakan untuk membangkitkan pertanyaan juga harus disesuaikan dengan bahasa yang digunakan. Otomatisasi pembangkitan pertanyaan untuk bahasa Indonesia juga sudah mulai berkembang sejak 2015. Hasil penelitian-penelitian tersebut perlu dianalisis agar dapat mengetahui kelebihan maupun kekurangan dari setiap metode yang pernah digunakan. Oleh karena itu, jurnal ini membahas tentang Systematic Literature Review (SLR) pembangkitan pertanyaan pada bahasa Indonesia. SLR yang dibangun ini dapat digunakan untuk bahan pertimbangan optimalisasi penelitian tentang pembangkitan pertanyaan menggunakan bahasa Indonesia di kemudian hari. Tahapan yang dilakukan dalam pembentukan SLR adalah perencanaan literature review, kemudian melakukan literature review dan terakhir adalah pelaporan hasil literature review. Pencarian pada google scholar menghasilkan 27 penelitian yang relevan dengan kata kunci. Penerapan kriteria inklusi dan eksklusi menghasilkan 15 penelitian yang relevan. Kemudian proses backward dan forward snowballing yang dilakukan menghasilkan 2 penelitian tambahan. Total penelitian yang dianalisis berjumlah 17 penelitian. Proses selanjutnya adalah penilaian kualitas penelitian. Hasil penilaian kualitas penelitian menunjukkan bahwa keseluruhan penelitian yang berjumlah 17 penelitian tersebut memiliki kualitas yang baik untuk dianalisis. Hasil analisis yang dilakukan menunjukkan bahwa penelitian awal terkait pembangkitan pertanyaan untuk bahasa Indonesia masih memiliki beberapa celah. Diantaranya terkait dataset yang belum memadai, model pertanyaan yang kurang beragam, belum adanya penanganan/preproses model kalimat unstructured, dan belum adanya pembangkitan pertanyaan yang berasal dari gabungan beberapa informasi. Abstract Research on automated question generation is constantly evolving. Various methods have been tried to be applied in various languages. Each language has different characteristics. Therefore, the method used to generate questions should be adapted based on the language. The automation of question generation for Indonesian has also begun to develop since 2015. The result of these studies need to be analyzed to find out the advantages and disadvantages of each method that has been used. Therefore, this journal discusses the Systematic Literature Review (SLR) for generating questions in Indonesian. The SLR that was built can be used for consideration of optimizing research on generating questions using Indonesian in the future. The steps taken in this SLR are planning analysis, then carrying out the analysis and finally reporting the analysis. A search on Google Scholar yielded 27 studies that were relevant to the keyword. The application of inclusion and exclusion criteria resulted in 15 relevant studies. Then the backward and forward snowballing processes carried out resulted in 2 additional studies. Total research analyzed amounted to 17 studies. The next process is the assessment of research quality. The results of the research quality assessment showed that the overall 17 studies had good quality for analysis. The results of the analysis carried out indicate that the initial research related to question generation for Indonesian still has some gaps. For examples about datasets, question models, handling unstructured sentence models, and generating questions from a combination of some information.

Pronomina (kata ganti) adalah jenis kata yang dapat dipakai untuk menggantikan posisi kata benda atau orang dalam suatu kalimat. Penggunaan pronomina akan mudah dipahami apabila serangkaian kalimat dibaca secara utuh. Namun jika rangkaian kalimat tersebut hanya dibaca pada kalimat-kalimat tertentu, maka akan sulit memahami kalimat yang memiliki pronomina. Pada pengolahan bahasa alamiah, diperlukan kejelasan konteks dari sebuah kalimat. Dalam konteks otomatisasi pengolahan bahasa alamiah, adanya pronomina dapat menyulitkan komputer untuk memahami kalimat tersebut. Oleh sebab itu, dalam pengolahan bahasa alamiah yang mengandung pronomina diperlukan pre proses berupa pengubahan pronomina ke dalam bentuk subjek atau objek asli yang dirujuk. Metode yang diusulkan untuk menyelesaikan permasalahan ini adalah pendekatan berbasis sintaktik. Pendekatan ini menitikberatkan pada struktur kata yang digunakan dan struktur komponen kata yang digunakan. Metode yang diusulkan memiliki 4 tahapan yakni pengumpulan data, pembangkitan aturan, otomatisasi pengenalan pronominal, dan terakhir adalah evaluasi. Metode yang diusulkan telah diujicobakan untuk mengenali adanya pronomina dari kalimat-kalimat pada materi Ilmu Pengetahuan Alam dan Ilmu Pengetahuan Sosial di jenj­­ang sekolah dasar. Hasil evaluasi menunjukkan bahwa metode yang diusulkan dapat digunakan untuk mengubah subjek yang berbentuk pronomina menjadi subjek atau objek asli yang dirujuk. Rata-rata akurasi yang didapatkan sebesar 81%. Akurasi tersebut didapatkan dari perbandingan antara jumlah kata ganti yang berhasil diidentifikasi subjeknya dengan keseluruhan data uji. Hasil dari penelitian ini dapat digunakan peneliti di bidang Natural Language Processing untuk melakukan praproses terhadap teks yang akan diolah. Abstract A pronoun is a word that can be used to replace a noun or person in a sentence. The use of pronouns will be easy to understand if a series of sentences is read in its entirety. However, if the sentence series is only read in specific sentences, it will be difficult to understand sentences with pronouns. In natural language processing, it is necessary to clarify the context of a sentence. In the context of automation of natural language processing, the existence of pronouns can make it difficult for computers to understand the sentence. Therefore, in processing natural language containing pronouns, it is necessary to pre-process in the form of converting pronouns into the form of the original subject or object referred to. The method proposed to solve this problem is a syntactic-based approach. This approach focuses on the structure of the words used and the word components used. The proposed method has 4 stages, namely data collection, rule generation, automation of pronoun recognition and the last is evaluation. The proposed method has been evaluated to identify the existence of pronouns from sentences in the Natural Sciences and Social Sciences material at the elementary school level. The evaluation results show that the proposed method can be used to change the subject in the form of a pronoun into the original subject or object referred to. The average accuracy obtained is 81%. The accuracy is obtained from the comparison between the number of pronouns that have been identified with the overall test data. Researchers in natural language processing can use the results of this study to pre-process their text.

People who have hearing loss (deafness) or speech impairment (hearing impairment) usually use sign language to communicate. One of the most basic and flexible sign languages ​​is the Alphabet Sign Language to spell out the words you want to pronounce. Sign language uses hand, finger, and face movements to speak the user's thoughts. However, for alphabetical sign language, facial expressions are not used but only gestures or symbols formed using fingers and hands. In fact, there are still many people who don't understand the meaning of sign language. The use of image classification can help people more easily learn and translate sign language. Image classification accuracy is the main problem in this case. This research conducted a comparison of image classification algorithms, namely Convolutional Neural Network (CNN) and Multilayer Perceptron (MLP) to recognize American Sign Language (ASL) except the letters "J" and "Z" because movement is required for both. This is done to see the effect of the convolution and pooling stages on CNN on the resulting accuracy value and F1 Score in the ASL dataset. Based on the comparison, the use of CNN which begins with Gaussian Low Pass Filtering preprocessing gets the best accuracy of 96.93% and F1 Score 96.97%

Background: The COVID-19 pandemic remains a problem in 2021. Health protocols are needed to prevent the spread, including wearing a face mask. Enforcing people to wear face masks is tiring. AI can be used to classify images for face mask detection. There are a lot of image classification algorithm for face mask detection, but there are still no studies that compare their performance. Objective: This study aims to compare the classification algorithms of classical machine learning. They are k-nearest neighbors (KNN), support vector machine (SVM), and a widely used deep learning algorithm for image classification which is convolutional neural network (CNN) for face masks detection. Methods: This study uses 5 and 3 cross-validation for assessing the performance of KNN, SVM, and CNN in face mask detection. Results: CNN has the best average performance with the accuracy of 0.9683 and average execution time of 2,507.802 seconds for classifying 3,725 faces with mask and 3,828 faces without mask images. Conclusion: For a large amount of image data, KNN and SVM can be used as temporary algorithms in face mask detection due to their faster execution times. At the same time, CNN can be trained to form a classification model. In this case, it is advisable to use CNN for classification because it has better performance than KNN and SVM. In the future, the classification model can be implemented for automatic alert system to detect and warn people who are not wearing face masks.

The increasing number of cars that have been released to the market makes it more difficult for buyer to choose the choice of car that fits with their desired criteria such as transmission, number of kilometers, fuel type, and the year the car was made. The method that is suitable in determining the criteria desired by the community is the K-Nearest Neighbors (KNN). This method is used to find the lowest distance from each data in a car with the criteria desired by the buyer. Euclidean, Manhattan, and Minkowski distance are used for measuring the distance. For supporting the selection of cars, we need an automatic data col-lection method by using web crawling in which the system can retrieve car data from several ecommerce websites. With the construction of the car search system, the system can help the buyer in choosing a car and Euclidean distance has the best accuracy of 94.40%.

Permainan batu, gunting, dan kertas sangat populer di seluruh dunia. Permainan ini biasanya dimainkan saat sedang berkumpul untuk mengundi ataupun hanya bermain untuk mengetahui yang menang dan yang kalah. Namun, perkembangan zaman dan teknologi mengakibatkan orang dapat berkumpul secara virtual. Untuk bisa melakukan permainan ini secara virtual, penelitian ini membuat model klasifikasi citra untuk membedakan objek tangan yang menunjuk batu, kertas, dan gunting. Performa metode klasifikasi merupakan hal yang harus diperhatikan dalam kasus ini. Salah satu metode klasifikasi citra yang populer adalah Convolutional Neural Network (CNN). CNN adalah salah satu jenis neural network yang biasa digunakan pada data klasifikasi citra. CNN terinspirasi dari jaringan syaraf manusia. Algoritma ini memiliki 3 tahapan yang dipakai, yaitu convolutional layer, pooling layer, dan fully connected layer. Uji coba 5-Fold cross validation klasifikasi objek tangan yang menunjuk citra batu, kertas, dan gunting menggunakan CNN pada penelitian ini menghasilkan rata-rata akurasi sebesar 97.66%.

Pada tahun 2021 pandemi Covid-19 masih menjadi masalah di dunia. Protokol kesehatan diperlukan untuk mencegah penyebaran Covid-19. Penggunaan masker wajah adalah salah satu protokol kesehatan yang umum digunakan. Pengecekan secara manual untuk mendeteksi wajah yang tidak menggunakan masker adalah pekerjaan yang lama dan melelahkan. Computer vision merupakan salah satu cabang ilmu komputer yang dapat digunakan untuk klasifikasi citra. Convolutional Neural Network (CNN) merupakan algoritma deep learning yang memiliki performa bagus dalam klasifikasi citra. Transfer learning merupakan metode terkini untuk mempercepat waktu training pada CNN dan untuk mendapatkan performa klasifikasi yang lebih baik. Penelitian ini melakukan klasifikasi citra wajah untuk membedakan orang menggunakan masker atau tidak dengan menggunakan CNN dan Transfer Learning. Arsitektur CNN yang digunakan dalam penelitian ini adalah MobileNetV2, VGG16, DenseNet201, dan Xception. Berdasarkan hasil uji coba menggunakan 5-cross validation, Xception memiliki akurasi terbaik yaitu 0.988 dengan waktu total komputasi training dan testing sebesar 18274 detik. MobileNetV2 memiliki waktu total komputasi tercepat yaitu 4081 detik dengan akurasi sebesar 0.981. Abstract In 2021 the Covid-19 pandemic is still a problem in the world. Therefore, health protocols are needed to prevent the spread of Covid-19. The use of face masks is one of the commonly used health protocols. However, manually checking to detect faces that are not wearing masks is a long and tiring job. Computer vision is a branch of computer science that can be used for image classification. Convolutional Neural Network (CNN) is a deep learning algorithm that has good performance in image classification. Transfer learning is the latest method to speed up CNN training and get better classification performance. This study performs facial image classification to distinguish people using masks or not by using CNN and Transfer Learning. The CNN architecture used in this research is MobileNetV2, VGG16, DenseNet201, and Xception. Based on the results of trials using 5-cross validation, Xception has the best accuracy of 0.988 with a total computation time of training and testing of 18274 seconds. MobileNetV2 has the fastest total computing time of 4081 seconds with an accuracy of 0.981.

Kondisi pandemi global Covid-19 yang muncul diakhir tahun 2019 telah menjadi permasalahan utama seluruh negara di dunia. Covid-19 merupakan virus yang menyerang organ paru-paru dan dapat mengakibatkan kematian. Pasien Covid-19 banyak yang telah dirawat di rumah sakit sehingga terdapat data citra chest X-ray paru-paru pasien yang terjangkit Covid-19. Saat ini sudah banyak peneltian yang melakukan klasifikasi citra chest X-ray menggunakan Convolutional Neural Network (CNN) untuk membedakan paru-paru sehat, terinfeksi covid-19, dan penyakit paru-paru lainnya, namun belum ada penelitian yang mencoba membandingkan performa algoritma CNN dan machine learning klasik seperti Support Vector Machine (SVM), dan K-Nearest Neighbor (KNN) untuk mengetahui gap performa dan waktu eksekusi yang dibutuhkan. Penelitian ini bertujuan untuk membandingkan performa dan waktu eksekusi algoritma klasifikasi K-Nearest Neighbors (KNN), Support Vector Machine (SVM), dan CNN untuk mendeteksi Covid-19 berdasarkan citra chest X-Ray. Berdasarkan hasil pengujian menggunakan 5 Cross Validation, CNN merupakan algoritma yang memiliki rata-rata performa terbaik yaitu akurasi 0,9591, precision 0,9592, recall 0,9591, dan F1 Score 0,959 dengan waktu eksekusi rata-rata sebesar 3102,562 detik.

Cuaca merupakan faktor penting yang dipertimbangkan untuk berbagai pengambilan keputusan. Klasifikasi cuaca manual oleh manusia membutuhkan waktu yang lama dan inkonsistensi. Computer vision adalah cabang ilmu yang digunakan komputer untuk mengenali atau melakukan klasifikasi citra. Hal ini dapat membantu pengembangan self autonomous machine agar tidak bergantung pada koneksi internet dan dapat melakukan kalkulasi sendiri secara real time. Terdapat beberapa algoritma klasifikasi citra populer yaitu K-Nearest Neighbors (KNN), Support Vector Machine (SVM), dan Convolutional Neural Network (CNN). KNN dan SVM merupakan algoritma klasifikasi dari Machine Learning sedangkan CNN merupakan algoritma klasifikasi dari Deep Neural Network. Penelitian ini bertujuan untuk membandingkan performa dari tiga algoritma tersebut sehingga diketahui berapa gap performa diantara ketiganya. Arsitektur uji coba yang dilakukan adalah menggunakan 5 cross validation. Beberapa parameter digunakan untuk mengkonfigurasikan algoritma KNN, SVM, dan CNN. Dari hasil uji coba yang dilakukan CNN memiliki performa terbaik dengan akurasi 0.942, precision 0.943, recall 0.942, dan F1 Score 0.942. Abstract Weather is an important factor that is considered for various decision making. Manual weather classification by humans is time consuming and inconsistent. Computer vision is a branch of science that computers use to recognize or classify images. This can help develop self-autonomous machines so that they are not dependent on an internet connection and can perform their own calculations in real time. There are several popular image classification algorithms, namely K-Nearest Neighbors (KNN), Support Vector Machine (SVM), and Convolutional Neural Network (CNN). KNN and SVM are Machine Learning classification algorithms, while CNN is a Deep Neural Networks classification algorithm. This study aims to compare the performance of that three algorithms so that the performance gap between the three is known. The test architecture is using 5 cross validation. Several parameters are used to configure the KNN, SVM, and CNN algorithms. From the test results conducted by CNN, it has the best performance with 0.942 accuracy, 0.943 precision, 0.942 recall, and F1 Score 0.942.

Data mining and process mining provide solutions for fraud detection. The automated methods based on the historical data, however, still need an improvement. In this regard, we propose a hybrid method between association rule learning and process mining. The process mining, in this case, inspects the event log. Through an expert verification, the itemset of the association rule learning is used to generate positive and negative rules applied for compliance checking towards the testing dataset. The result then shows that the hybrid method has less false discovery rate and provides higher accuracy compared to that of the process-mining method in which the optimum accuracy lies in certain threshold of confidence level. [ABSTRACT FROM AUTHOR]