Jurnal RESTI (Rekayasa Sistem dan Teknologi Informasi)
Not a member yet
1071 research outputs found
Sort by
On the Neural Network Solution of One-Dimensional Wave Problem
Get PDFArtificial neural network has become an emerging popular method to handle various problems, especially in case where it has deep multiple neural layers. In this study, we use a deep artificial neural network model to solve one-dimensional wave equation, without any external datasets. Different type of boundary conditions, i.e., Dirichlet, Neumann, and Robin, are used. We analyze the model learning capabilities in a set of settings, such as data setup and the model width and depth. We also present some discussions of advantages and disadvantages of the model in comparison with other matured existing techniques to solve wave equation. Artificial neural network has become an emerging popular method to handle various problems, especially in case where it has deep multiple neural layers. In this study, we use a deep artificial neural network model to solve one-dimensional wave equation, without any external datasets. Different type of boundary conditions, i.e., Dirichlet, Neumann, and Robin, are used. We analyze the model learning capabilities in a set of settings, such as data setup and the model width and depth. We also present some discussions of advantages and disadvantages of the model in comparison with other matured existing techniques to solve wave equation.  
Telegram Bot as a Data Collection Tool for Progress Reports in Area Mapping Progress Monitoring System
Get PDFIn the Area Mapping Project (Preparation for the 2020 Population Census), there is monitoring and collecting data process on the progress of activities in the field. There are weaknesses in data collection on the progress of activities in the field; The manual recapitulation of progress reporting makes the progress data not displayed in real-time, the SMS Gateway is often interrupted, and progress data collection through the monitoring website is less effective if reported directly by field officers. Telegram Bot is used as a data collection tool for field progress reports on Area Mapping activities to overcome these weaknesses. This study aims to prove that Telegram Bot can be used as a real-time data collection tool, has good performance, and is acceptable to users. Telegram Bot is integrated with Monitoring Website into one system and database in this research. This system uses PHP, Yii2, and MySQL, and communication between the web server and Telegram Server uses the webhook method. Based on the Black Box test results, all functions in this system run as expected. The average bot response time was 7.72 seconds for images and 2.25 seconds for text data in the performance test. In the usability test, Telegram Bot obtained a SUS Score of 66 and an NPS of 12.195. These results show that Telegram Bot can be used as a real-time data collection tool, has good performance, and is well accepted by users.In the Area Mapping Project (Preparation for the 2020 Population Census), there is monitoring and collecting data process on the progress of activities in the field. There are weaknesses in data collection on the progress of activities in the field; The manual recapitulation of progress reporting makes the progress data not displayed in real-time, the SMS Gateway is often interrupted, and progress data collection through the monitoring website is less effective if reported directly by field officers. Telegram Bot is used as a data collection tool for field progress reports on Area Mapping activities to overcome these weaknesses. This study aims to prove that Telegram Bot can be used as a real-time data collection tool, has good performance, and is acceptable to users. Telegram Bot is integrated with Monitoring Website into one system and database in this research. This system uses PHP, Yii2, and MySQL, and communication between the web server and Telegram Server uses the webhook method. Based on the Black Box test results, all functions in this system run as expected. The average bot response time was 7.72 seconds for images and 2.25 seconds for text data in the performance test. In the usability test, Telegram Bot obtained a SUS Score of 66 and an NPS of 12.195. These results show that Telegram Bot can be used as a real-time data collection tool, has good performance, and is well accepted by users
Adaptive Streaming Server dengan FFMPEG dan Golang
Get PDFQuality of experience (QoE) when accessing video streaming becomes a challenge in varieties of network bandwidth/speed. Adaptive streaming becomes an answer to gain good QoE. An architecture system of the adaptive streaming server with Dynamic Adaptive Streaming over HTTP (DASH) was proposed. The system was consists of two services e.g transcoding and streaming. Transcoding service encodes an audio file, multi-bitrates video files, and manifest.mpd files. Streaming service serves client streaming requests that appropriate to client network profiles. The system is built using the Golang programming environment and FFMPEG. Transcoding service has some execution mode (serial and concurrent) and passing mode (1 pass and 2 passes). The transcoding service test results show that concurrent execution is faster 11,5% than the serial execution and transcoding using 1 pass is faster 46,95% than 2 passes but the bitrate of output video lower than the determinate bitrate parameter. The streaming service has a good QoE. In the 5 scenarios, buffer level=0 happens 5 times, and its total duration is 64 seconds. Buffer level=0 happens when extreme changes happen in network speed from fast to too slow.
Kualitas pengalaman pengguna (QoE) dalam mengakses video streaming menjadi salah satu tantangan di tengah bervariasinya bandwidth/kecepatan jaringan. Adaptive streaming merupakan salah satu jawaban dalam mencapai QoE yang baik. Sebuah rancangan sistem adaptive streaming server dengan konsep Dynamic Adaptive Streaming over HTTP (DASH) disusun untuk menjawab masalah ini. Sistem terdiri dari dua layanan yaitu transcoding dan streaming. Layanan transcoding bertugas menyiapkan file audio, file-file video multi bitrate dan file manifest.mpd. Layanan streaming bertugas melayani permintaan streaming yang telah disesuaikan dengan kondisi jaringan klien. Sistem yang dibangun menggunakan FFMPEG dan lingkungan pemrograman Golang. Layanan transcoding dilengkapi beberapa kemampuan eksekusi (serial dan konkuren) dan passing mode (1 pass & 2 pass). Pada ujicoba layanan transcoding diperoleh bahwa eksekusi secara konkuren lebih cepat 11,5% dibandingkan secara serial. Transcoding dengan 1 pass dibandingkan 2 pass memiliki durasi lebih cepat 46,95% tetapi bitrate video hasil transcoding jauh di bawah parameter bitrate yang ditentukan. QoE layanan streaming cukup baik. Dari 5 skenario terjadi 5 kali buffer level=0 dengan total durasinya 64 detik. Buffer level=0 terjadi ketika ada perubahan yang ekstrem terhadap kecepatan jaringan dari yang semula cepat ke sangat lambat
Implementation of Verification and Matching E-KTP with Faster R-CNN and ORB
Get PDFneeds a solid validation that has verification and matching uploaded images. To solve this problem, this paper implementing a detection model using Faster R-CNN and a matching method using ORB (Oriented FAST and Rotated BRIEF) and KNN-BFM (K-Nearest Neighbor Brute Force Matcher). The goal of the implementations is to reach both an 80% mark of accuracy and prove matching using ORB only can be a replaced OCR technique. The implementation accuracy results in the detection model reach mAP (Mean Average Precision) of 94%. But, the matching process only achieves an accuracy of 43,46%. The matching process using only image feature matching underperforms the previous OCR technique but improves processing time from 4510ms to 60m). Image matching accuracy has proven to increase by using a high-quality dan high quantity dataset, extracting features on the important area of EKTP card images.needs a solid validation that has verification and matching uploaded images. To solve this problem, this paper implementing a detection model using Faster R-CNN and a matching method using ORB (Oriented FAST and Rotated BRIEF) and KNN-BFM (K-Nearest Neighbor Brute Force Matcher). The goal of the implementations is to reach both an 80% mark of accuracy and prove matching using ORB only can be a replaced OCR technique. The implementation accuracy results in the detection model reach mAP (Mean Average Precision) of 94%. But, the matching process only achieves an accuracy of 43,46%. The matching process using only image feature matching underperforms the previous OCR technique but improves processing time from 4510ms to 60m). Image matching accuracy has proven to increase by using a high-quality dan high quantity dataset, extracting features on the important area of EKTP card images
Prototype Sistem Pakan Ikan dan Pemantauan PH Berbasis Android dengan Metode PLC
Get PDFAbstract
During the pandemic of covid-19, betta fish cultivation is one of the income alternatives. It makes the cultivation of betta fish is increasing. However, not all cultivators can successfully cultivate betta fish. Water quality and fish feed must be maintained in order for betta fish can grow perfectly and attractive. The problems that often encountered to the betta fish cultivator is about feeding and monitoring pH levels, which is still manual. This is can failure in betta fish cultivation of the cultivators are not disciplined. To minimize it, then will be made “Prototype Fish Feeding System and PH Monitoring Based Android”. This research is used PLC method: (a) Requirements Analysis, (b) Design, and (c) Implementation. Divided into 2 functions: Automatic Betta Fish Feeding System Based Android Function using Node MCU ESP8266 and Monitoring PH Levels Function using Arduino Uno. The result from this research is the user can control the open and close feed on the servo through an android application, the system can provide a distance of the feed supply to a user in real-time by notifications, the LCD always display pH value, the buzzer provides a sound when the water quality has reached the acid or alkaline index.
Di masa pandemi covid-19 saat ini, budidaya ikan cupang menjadi salah satu alternatif sumber penghasilan yang menjanjikan. Sehingga pembudidayaan ikan cupang menjadi semakin meningkat. Namun, tidak semua pembudidaya dapat berhasil membudidayakan ikan cupang. Agar ikan cupang dapat bertumbuh dengan sempurna dan menarik, kualitas air dan pakan ikan harus dijaga. Permasalahan yang sering dijumpai oleh pemelihara ikan cupang yaitu dalam pemberian pakan dan pemantauan kualitas pH air yang kurang efisien, masih dilakukan secara manual. Hal ini dapat menyebabkan kegagalan dalam budidaya ikan cupang apabila kurangnya kedisiplinan dari pembudidaya ikan cupang. Untuk meminimalisir hal itu terjadi, dibuatlah “Prototype Sistem Pakan Ikan dan Pemantauan PH Berbasis Android”. Metode yang digunakan adalah metode PLC dengan tahap, (a) Penjabaran spesifikasi kebutuhan, (b) Perancangan Desain Sistem, dan (c) Implementasi. Penelitian ini terbagi menjadi 2 fungsi yaitu, Fungsi Pemberi Pakan Ikan Berbasis Android dengan menggunakan mikrokontroler Node MCU ESP8266 dan Fungsi Pemantauan Kadar pH dengan menggunakan mikrokontroler Arduino. Hasil akhir dari penelitian ini adalah Pengguna dapat mengendalikan buka dan tutup pakan pada servo melalui aplikasi yang telah terpasang pada smartphone Android, Sistem dapat memberikan notifikasi jarak persediaan pakan kepada pengguna secara real time, LCD menampilkan nilai PH secara real time, Buzzer memberikan peringatan berupa bunyi ketika kualitas nilai air sudah mencapai indeks asam atau basa. Sistem ini baik apabila digunakan untuk pembudidaya maupun penghobi ikan cupang
Implementasi Sistem Otentikasi Dokumen Berbasis Quick Response (QR) Code dan Digital Signature
Get PDFThe authenticity and integrity of documents are essential in data exchange and communication. Digital documents must be verifiable for their authenticity and integrity by all parties that use the documents. Generally, digital documents can be authenticated by using digital signatures. This study aims to implement a document authentication system based on Quick Response (QR) code and digital signature. As the case study, the document authentication system is implemented to generate digital signatures for student’s certificate documents. Furthermore, the system can also verify the authenticity of the certificate documents. Creating a digital signature requires a hash function algorithm for generating the message digest of the document. In addition, an algorithm to generate the public key and the private key used in the encryption/decryption of the message digest is also needed. The hash function utilized in this study is the Secure Hash Algorithm-256 (SHA-256), while the algorithm used for encryption/decryption is the Rivest-Shamir-Adleman (RSA) algorithm. The system is evaluated by verifying 30 student certificate documents, of which 15 of them were certificates with QR code signature generated by the system and the other 15 were certificates with QR code signature generated using a random QR code generator. The system’s testing results demonstrate that the system can ensure the authenticity and integrity of the signed certificate documents to prevent document falsification. All documents that contain random QR codes were correctly identified as false documents.
Keaslian dan integritas dari sebuah dokumen merupakan hal yang esensial dalam pertukaran data pada aplikasi teknologi informasi dan komunikasi. Oleh karena itu, suatu dokumen digital harus dapat diverifikasi keasliannya oleh pihak-pihak yang menggunakan dokumen tersebut agar dapat diketahui apakah dokumen tersebut otentik atau tidak. Secara umum, dokumen digital dapat dijaga keasliannya menggunakan metode atau teknik dalam kriptografi, yaitu tanda tangan digital (digital signature). Penelitian ini bertujuan untuk mengimplementasikan suatu sistem otentikasi dokumen berbasis Quick Response (QR) code dan digital signature. Sistem otentikasi dokumen yang diimplementasikan digunakan untuk menguji keaslian pada dokumen sertifikat mahasiswa sebagai contoh kasusnya. Dalam pembuatan digital signature dibutuhkan suatu fungsi hash untuk membuat message digest dari dokumen yang akan ditandatangani. Selain itu, suatu algoritma untuk melakukan enkripsi/dekripsi serta membangkitkan kunci publik/kunci privat juga dibutuhkan dalam proses pembangkitan tanda tangan digital. Penelitian ini menggunakan Secure Hash Algorithm-256 (SHA-256) sebagai fungsi hash dan algoritma Rivest-Shamir-Adleman (RSA) untuk proses enkripsi/dekripsi dan pembangkitan kunci publik/kunci privat. Pengujian sistem dilakukan terhadap 30 sertifikat, di mana 15 sertifikat mencantumkan QR code yang dihasilkan oleh sistem sebagai signature-nya dan 15 sertifikat lainnya memuat signature QR code yang dibuat menggunakan QR code generator lain. Dari pengujian sistem didapatkan hasil bahwa implementasi sistem otentikasi dokumen berbasis QR code dan digital signature ini dapat memastikan keaslian dan integritas dokumen sehingga mencegah pemalsuan dokumen. Semua dokumen yang mencantumkan QR code yang tidak dibuat atau di-generate oleh sistem ini berhasil diidentifikasi sebagai dokumen palsu
Perbandingan Support Vector Machine dan Modified Balanced Random Forest dalam Deteksi Pasien Penyakit Diabetes
Get PDFDiabetes (diabetes) was a metabolic disorder caused by high levels of sugar in the blood caused by disorders of the pancreas and insulin. According to data from the Ministry of Health of the Republic of Indonesia, Diabetes was the third-largest cause of death in Indonesia with a percentage of 6.7%. The high rate of death from diabetes encouraged this study, with the aim of early detection. This research used a Machine Learning approach to classify the data. In this paper, a comparison of Support Vector Machine (SVM) and Modified Balanced Random Forest (MBRF) was discussed for classifying diabetes patient data. Both methods were chosen because it was proven in previous studies to get high accuracy, so that the two methods are compared to find the best classification model. Several preprocessing methods were used to prepare the data for the classification process. The entire combination of preprocessing steps will be carried out on the two classification methods to produce the same dataset. The evaluation was carried out using the Confusion Matrix method. Based on the experimental results in the process of testing the system being built, the maximum performance results were 87.94% using SVM and 97.8% using MBRF.Diabetes (kencing manis) merupakan suatu kelainan metabolik yang disebabkan oleh tingginya tingkat kandungan gula dalam darah yang diakibatkan oleh gangguan pada pankreas dan insulin. Menurut data dari Kementrian Kesehatan Republik Indonesia, Diabetes merupakan penyebab kematian terbesar nomor 3 di Indonesia dengan persentase sebesar 6,7%. Tingginya tingkat kematian akibat diabetes mendorong dilakukan penelitian ini, dengan tujuan untuk deteksi dini. Pada penilitian ini akan menggunakan pendekatan Machine Learning untuk melakukan klasifikasi datanya. Dalam makalah ini, dibahas perbandingan Suport Vector Machine (SVM) dan Modified Balanced Random Forest (MBRF) untuk melakukan klasifikasi data pasien diabetes. Kedua metode dipilih karena terbukti pada penelitian sebelumnya mendapatkan akurasi yang tinggi, sehingga kedua metode tersebut dibandingkan untuk mencari model klasifikasi yang terbaik. Beberapa metode preprocessing dilakukan untuk mempersiapkan data agar dapat di lakukan proses klasifikasi. Seluruh kombinasi tahapan dari preprocessing akan dilakukan terhadap kedua metode klasifikasi untuk menghasilkan dataset yang sama juga. Evaluasi dilakukan menggunakan metode Confusion Matrix Berdasarkan hasil eksperimen dalam proses pengujian sistem yang dibangun, diperoleh hasil performasi maksimum 87,94% dengan menggunakan SVM dan 97,8% dengan menggunakan MBRF
Penerapan Convolutional Neural Network Deep Learning dalam Pendeteksian Citra Biji Jagung Kering
Get PDFCorn kernels detection can be implemented in industry area. This can be implemented in the selection and packaging the corn kernels before it is distributed. This technique can be implemented in the selection and packaging machine to detect corn kernels accurately. Corn kernel images was used before it is implemented in real-time. The objective of this research was corn kernel detection using Convolutional Neural Network (CNN) deep learning. This technique consists of 3 main stages, the first preprocessing or normalizing the input of corn kernels image data by wrapping and cropping, both modeling and training the system, and testing. The experiment used CNN method to classify images of dry corn kernels and to determine the accuracy value. This research used 20 dry corn kernels images as testing from 80 dry corn kernels images which used in training dataset. The accuracy of detection was dependent from the size of image and position when the image was taken. The accuracy is around 80% - 100% by using 7 convolutional layers and the average of accuracy for testing data was 0,90296. The convolutional layer which implemented in CNN has the strength to detect features in the input image.
Pendeteksian bji jagung kering dapat diimplementasikan pada dunia industri. Khususnya ketika pemilahan dan pembungkusan biji jagung kering dilakukan sebelum dipasarkan. Saat ini pemilahan dan pembungkusan belum mengimplementasikan deteksi biji jagung kering sehingga terkadang di dalam kemasan biji jagung kering sering terdapat biji lainnya. Metode pendeteksian ini dapat diintegrasikan dengan mesin pemilah dan pembungkus di industri biji-bijian salah satunya untuk mendeteksi biji jagung kering. Untuk mendapatkan proses pendeteksian yang akurat, citra biji jagung kering digunakan sebagai data pada metode deep learning sebelum implementasi secara real-time. Tujuan dari penelitian ini adalah mendeteksi citra biji jagung kering dengan menerapkan metode Convolutional Neural Network (CNN) deep learning. Teknik ini terdiri dari 3 tahap utama, pertama preprocessing atau menormalkan data input citra biji jagung dengan melakukan wrapping dan cropping. Kedua, pembentukan model dan pelatihan sistem, yang terakhir adalah melakukan untuk pengujian sistem. Penelitian menggunakan CNN untuk mengenali citra biji jagung kering dan menentukan nilai akurasinya. Pada penelitian ini digunakan 20 citra biji jagung yang digunakan sebagai testing data dari 80 citra biji jagung yang digunakan pada training dataset. Nilai akurasi pendeteksian biji jagung kering dipengaruhi oleh ukuran citra dan posisi pengambilan citra dari kamera smartphone. Penggunaan 7 convolutional layer memberikan nilai akurasi berkisar antara 80% - 100% sehingga nilai rata-rata akurasi testing data sebesar 0,90296. Penggunaan convolutional layer mampu mendeteksi kekuatan bentuk dari suatu citra
Studi Komparasi Model Klasifikasi Berbasis Pembelajaran Mesin untuk Sistem Rekomendasi Program Studi
Get PDFSelecting a major can be quite difficult for prospective college students. The choice may have an effect not only on their academic life, but also on their career path. Due to some restrictions as the impact of the COVID-19 pandemic, universities must find novel ways to reach prospective students and assist them in choosing their majors, one of which is a college major recommendation system. This system can assist prospective students in determining the most appropriate majors for them based on data from the current students. Unlike other existing systems that employ either a rule-based or fuzzy model, this study employs a machine learning approach using data from undergraduate students at Universitas Islam Indonesia. This paper aims to compare several clustering models (i.e., K-means, Agglomerative, Birch, and DBSCAN) for the purpose of categorizing current students, to which the results will be used for classification purposes using various approaches (i.e., single stage vs. multistage), algorithms (i.e., multinomial logistic regression, random forest, and support vector machine), and scenarios (i.e., with or without GPA-based label). Our findings indicate that the K-means model outperformed all other clustering models and that the single stage with random forest classification model performed the best across all scenarios.
Pemilihan program studi di perguruan tinggi adalah salah satu tantangan bagi calon mahasiswa baru. Pilihan yang diambil dapat berdampak pada kehidupan akademis dan jalur karier mereka. Karena beberapa pembatasan sebagai dampak dari pandemi Covid-19, perguruan tinggi harus menemukan cara inovatif untuk menjangkau calon mahasiswa baru dan membantu mereka dalam memilih program studi, salah satunya dalam bentuk sistem rekomendasi program studi. Sistem ini dapat membantu calon mahasiswa baru dalam menentukan program studi yang paling sesuai untuk mereka berdasarkan data dari mahasiswa aktif di masing-masing program studi. Tidak seperti sistem lain yang menggunakan model berbasis aturan atau fuzi, penelitian ini menggunakan pembelajaran mesin (machine learning) dengan data dari mahasiswa program sarjana di Universitas Islam Indonesia untuk membangun sistem rekomendasi tersebut. Penelitian ini bertujuan untuk membandingkan beberapa model klastering (K-means, Agglomerative, Birch, dan DBSCAN) untuk mengkategorikan mahasiswa aktif yang hasilnya akan digunakan untuk proses klasifikasi menggunakan berbagai metode (single stage vs. multistage), algoritme (multinomial logistic regression, random forest, dan support vector machine), dan skenario (dengan atau tanpa label berbasis IPK). Temuan dari penelitian ini menunjukkan bahwa model K-means mengungguli semua model klastering lainnya sementara model klasifikasi single stage dengan random forest memiliki kinerja terbaik di semua skenario
Perbandingan Metode CBR dan Dempster-Shafer pada Sistem Pakar Terintegrasi Layanan Kesehatan
Get PDFInfectious disease is a very dangerous disease with a high mortality rate. Delays in handling the spread of an infectious disease can be minimized using an expert system. This study uses an expert system as a disease consulting service that is integrated with the health care system. Integration with the health care system is used for the knowledge acquisition process. The knowledge base on the expert system uses patient medical record data obtained through the health care system. The expert system can diagnose infectious diseases of sore throat (Pharyngitis), diphtheria, dengue fever, Typhoid fever, tuberculosis, and leprosy. The knowledge acquisition process produces 43 symptoms. These symptoms are used to diagnose new cases using Case-Based Reasoning (CBR) and Dempster-Shafer methods. In the CBR method, the similarity measurement process is determined by comparing the K-Nearest Neighbor, Minkowski Distance, and 3W-Jaccard similarity measurement methods. The expert system obtains accuracy values for the CBR K-Nearest Neighbor, CBR Minkowski Distance, and CBR 3W-Jaccard methods at a threshold of 70%, respectively 65.71%, 80%, and 85.71%. The average length of retrieve time required for each similarity method is 0.083s, 0.107s, and 6.325s, respectively. While the diagnosis of disease with Dempster-Shafer gets an accuracy value of 88.57%.
Penyakit menular merupakan penyakit yang sangat berbahaya dengan angka kematian cukup tinggi. Keterlambatan penanganan akan terjangkitnya suatu penyakit menular dapat diminimalisir menggunakan sistem pakar. Penelitian ini menggunakan sistem pakar sebagai layanan konsultasi penyakit yang diintegrasikan dengan sistem layanan kesehatan. Integrasi dengan sistem layanan kesehatan digunakan untuk proses akuisisi pengetahuan. Knowledge base pada sistem pakar menggunakan data rekam medis pasien yang didapatkan melalui sistem layanan kesehatan. Sistem pakar mampu mendiagnosis penyakit menular radang tenggorokan (Pharyngitis), difteri, demam berdarah, demam thypoid, tuberkulosis, dan kusta. Proses akuisisi pengetahuan menghasilkan 43 gejala. Gejala tersebut digunakan untuk mendiagnosis kasus baru dengan metode Case-Based Reasoning (CBR) dan Dempster-Shafer. Pada metode CBR proses pengukuran similaritas ditentukan dengan membandingkan metode pengukuran similaritas K-Nearest Neighbor, Minkowski Distance, dan 3W-Jaccard. Sistem pakar mendapatkan nilai akurasi untuk metode CBR K-Nearest Neighbor, CBR Minkowski Distance, dan CBR 3W-Jaccard pada threshold ≥ 70%, masing-masing 65.71%, 80%, dan 85.71%. Rata-rata lama waktu retrieve yang dibutuhkan setiap metode similaritas masing-masing 0.083s, 0.107s, dan 6.325s sedangkan diagnosis penyakit dengan Dempster-Shafer mendapatkan nilai akurasi sebesar 88.57%