Klasifikasi teks saat ini telah menjadi sebuah bidang yang banyak diteliti, khususnya terkait Natural Language Processing (NLP). Terdapat banyak metode yang dapat dimanfaatkan untuk melakukan klasifikasi teks, salah satunya adalah metode deep learning. RNN, CNN, dan LSTM merupakan beberapa metode deep learning yang umum digunakan untuk mengklasifikasikan teks. Makalah ini bertujuan menganalisis penerapan kombinasi dua buah metode deep learning, yaitu CNN dan LSTM (C-LSTM). Kombinasi kedua metode tersebut dimanfaatkan untuk melakukan klasifikasi teks berita bahasa Indonesia. Data yang digunakan adalah teks berita bahasa Indonesia yang dikumpulkan dari portal-portal berita berbahasa Indonesia. Data yang dikumpulkan dikelompokkan menjadi tiga kategori berita berdasarkan lingkupnya, yaitu “Nasional”, “Internasional”, dan “Regional”. Dalam makalah ini dilakukan eksperimen pada tiga buah variabel penelitian, yaitu jumlah dokumen, ukuran batch, dan nilai learning rate dari C-LSTM yang dibangun. Hasil eksperimen menunjukkan bahwa nilai F1-score yang diperoleh dari hasil klasifikasi menggunakan metode C-LSTM adalah sebesar 93,27%. Nilai F1-score yang dihasilkan oleh metode C-LSTM lebih besar dibandingkan dengan CNN, dengan nilai 89,85%, dan LSTM, dengan nilai 90,87%. Dengan demikian, dapat disimpulkan bahwa kombinasi dua metode deep learning, yaitu CNN dan LSTM (C-LSTM),memiliki kinerja yang lebih baik dibandingkan dengan CNN dan LSTM.
Short Message Service (SMS) is one of the features on a mobile phone. This feature is widely used because it is easy to use and does not require the latest telecommunications network connection. Short messages in the form of SMS consists of 140 characters at the most. The message is sent through infrastructure in telecommunications providers. Using this process, there is a possibility that the sent message is leaked. Therefore, data encryption is required to maintain the message confidentiality. Unfortunately, encryption mechanism uses cipher to encrypt data, which causes another problem. The type of cipher is symmetric or asymmetric, and both cipher mechanism will increase the length of the sent messages. In this paper, One Time Pad encryption method and LZW compression method is used to optimize the message length.
Liver is one of the most important organs in the human body. One of the dangerous diseases of the liver is tumor. In the CT scan image, the tumor has different texture, color, shape, and position, according to patient's condition. In this study, a tumor detection was carried out by tree stages: firstly some steps of preprocessing, such as filtering, edge detection, and erotion; secondly, finding the liver among organs in abdomen using segmentation and checking the liver position in the right abdomen; and thirdly performing the tumor detection in the liver using graph cut and push relabel algorithm. Usually, segmentation using graph cut needs two interactive inputs, namely sample of object area and sample of background area. In this paper, the interactive inputs on graph cut were replaced by deviation standard calculation. Testing using three sets of CT image and the ground truth produces average of the dice similarity coefficient (DSC), volumetric overlap error (VOE), and absolute volume difference (AVD) parameters of 78.15%, 25.72%, 19.30%, respectively. Furthermore, volume of liver tumor is approximated by utilizing area of tumor in each slice of CT image, then displayed in 3D view. Intisari-Liver atau hati merupakan salah satu organ penting di tubuh manusia. Salah satu penyakit berbahaya pada hati adalah tumor. Pada citra hasil CT scan, tumor mempunyai perbedaan tekstur, warna, bentuk, dan posisi yang terkait kondisi pasien. Pada makalah ini, deteksi tumor dilakukan melalui tiga tahap, pertama dengan praproses, di antaranya menggunakan penapisan, deteksi garis, dan erosi; kedua menemukan hati di antara berberapa organ dalam rongga perut dengan segmentasi dan deteksi posisi hati yang berada di sebelah kanan; dan ketiga adalah deteksi tumor pada hati yang telah dipisahkan dengan metode graph cut dan algoritme push relabel. Biasanya, segmentasi dengan graph cut memerlukan masukan interaktif berupa sampel area tumor (objek) dan sampel area latar belakang (background). Pada makalah ini, masukan interaktif tersebut digantikan dengan penggunaan parameter standard deviasi. Pengujian terhadap tiga set citra CT scan yang memiliki ground truth dari Kompetisi Segmentasi Tumor Hati oleh Miccai 2017 menghasilkan nilai rata-rata Dice Similarity Coefficient (DSC), Volumetric Overlap Error (VOE), dan Absolute Volume Difference (AVD) berturut-turut sebesar 78,15%, 25,72%, dan 19,30%. Selain itu, juga dilakukan penaksiran volume tumor hati dengan memanfaatkan luas kepingan tumor dari tiap potongan citra serta menampilkannya secara 3D.
Controlling each member of the soldiers to carry out battle with Non-Playable Characters (NPC) is one of the secrets to winning Real-Time Strategy games. The game could be more complicated and offer a more engaging experience if every NPC acts like humans rather than machines with patterned behavior. Like people during a war, each army member's command requires rapid reflexes and direction to strike or evade attacks. An intelligent opponent based on ANN as NPC can react quickly to their opponents. The accuracy of ANN could be enhanced by weight modifications using a Genetic Algorithm (GA). The crossover and mutation rates significantly impact GA's performance as an ANN setup. This research aims to find the best crossover and mutation rates in GA as a weight adjustment in ANN. Experiments were conducted using an RTS game simulator using 20 scenarios on a maximum of 4000 iterations. The initial setup of each troop is random, with a seven-unit type available. In this research, the troops won because their men were subjected to fewer attacks than the opposing forces. The GA optimal crossover and mutation rates are determined using troop victories as a baseline. According to the findings, the best crossover rate for GA as an ANN weight adjustment is 0.6, whereas the specific mutation rate is 0.09. The crossover rate of 0.6 has the highest average win value and tends to increase every generation. As for the mutation rate of 0.09, it has the highest average win value. Thus, this preliminary study can develop NPC more humanly.
The growth of the human population and technology has led to a rapid increase in electrical energy consumption. Excess electrical energy would be detrimental to the provider, whereas providing less would be detrimental to the consumers. One method for reducing these losses is to forecast the amount of electrical energy that must be available to meet demand. Prediction results can help with three types of decisions, depending on the prediction period: operational decisions (short-term), tactical decisions (medium-term), and strategic decisions (long-term). Short-term forecasts are less relevant given the urgency of the situation. This study aims to help electricity providers to make decisions by making medium and long-term predictions using the Auto-Regressive Integrated Moving Average (ARIMA) method. In the best order determination experiment, ARIMA (8,2,0) was found to be the best model with the smallest error. ARIMA (8,2,0) has an average percentage error of 5.3 percent based on the overall prediction results. There is no linearity between accuracy and prediction period in the prediction period experiment. According to the experimental results, the highest accuracy is obtained in the medium term (monthly) with a value of RMSE 753,983.98. As a result, based on the time period, ARIMA is the best for tactical decisions (medium-term) regarding electrical energy consumption.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.
hi@scite.ai
10624 S. Eastern Ave., Ste. A-614
Henderson, NV 89052, USA
Copyright © 2024 scite LLC. All rights reserved.
Made with 💙 for researchers
Part of the Research Solutions Family.