Ths paper aims to find similarity of Hepatitis B virus (HBV) and Hepatocelluler Carcinoma (HCC) DNA sequences. It is very important in bioformatics task. The similarity of sequence allignments indicates that they have similarity of chemical and physical properties. Mutation of the virus DNA in X region has potential role in HCC. It is observed using pairwise sequence alignment of genotype-A in HBV. The complexity of DNA sequence using dynamic programming, Needleman-Wunsch algorithm, is very high. Therefore, it is purpose to modifiy the method of Needleman Wunsch algorithm for optimum global DNA sequence alignment. The main idea is to optimize filling matrix and backtracking proccess of DNA components.This method can also solve various length of the both sequence alignment.This research is applied to DNA sequence of 858 hepatitis B virus and 12 carcinoma patient, so that there are 10,296 pairwis of sequences. They are aligned globally using the purposed method and as a result, it is achieved high similarity of 96.547% and validity of 99.854%. Furhthermore, this method has reduced the complexity of original Needleman-Wunsch algorithm The reduction of computational time is as 34.6% and space complexity is as 42.52%.

<p class="Abstrak"><em>Coronavirus</em> merupakan salah satu parasit yang menyerang sistem pernapasan manusia. Peningkatan kasus <em>coronavirus</em> berlangsung sangat cepat dan menyebar ke berbagai negara. Oleh karena itu, World Health Organization (WHO) menetapkan <em>Coronavirus</em> sebagai pandemi. Hal ini mengakibatkan seluruh kegiatan yang sebelumnya tatap muka atau luar jaringan (luring) menjadi dalam jaringan (daring), termasuk kegiatan belajar mengajar. Dengan ditetapkannya pembelajaran secara daring menyebabkan adanya opini yang bersifat pro dan kontra dari berbagai kalangan masyarakat. Opini tersebut akan digunakan dalam penelitian ini dan akan diolah terlebih dahulu dalam tahap <em>preprocessing</em>. Metode yang digunakan dalam penelitian ini adalah <em>Longest Common Subsequences</em> (LCS) dan <em>Support Vector Machine</em> (SVM)  dengan data sebesar 500 yang terbagi menjadi 250 data berlabel positif dan 250 data berlabel negatif. Dari 500 data tersebut dibagi menjadi 450 data untuk data latih dan 50 data untuk data uji. Dengan menggunakan metode <em>Longest Common Subsequences</em> untuk perbaikan kata dan metode <em>Support Vector Machine</em> untuk klasifikasi dengan nilai parameter terbaik yaitu <em>learning rate</em> (γ) = 0,0001, <em>lambda</em> (λ) = 0,1, <em>complexity</em> (C) = 0,001, <em>epsilon</em> (ϵ) = 0,0001 dan iterasi maksimum = 50 dapat menghasilkan nilai rata-rata hasil evaluasi yaitu <em>precision</em> = 0,5653, <em>recall</em> = 0,948, <em>f-measure</em> = 0,7047 dan <em>accuracy</em> = 0,598. Hasil pengujian tersebut mununjukkan bahwa dengan menambahkan metode <em>Longest Common Subsequences</em> untuk perbaikan kata dapat meningkatkan tingkat akurasi yang sebelumnya hanya 0,59 menjadi 0,598.</p><p class="Abstrak"> </p><p class="Abstrak"><strong>Abstract</strong></p><p class="Abstrak"> <em>Coronavirus is a parasite that attacks the human respiratory system. The increase incases coronavirus took place very fast and spread to various countries. Therefore, the World Health Organization (WHO) has designated Coronavirus as a pandemic. This results in all activities that were previously face-to-face or offline (offline) becoming online (online), including teaching and learning activities. With the establishment of online learning, there are pro and contra opinions from various circles of society. This opinion will be used in this research and will be processed first in the stage preprocessing. The method used in this research is Longest Common Subsequences (LCS) and Support Vector Machine (SVM) with 500 data divided into 250 data labeled positive and 250 data labeled negative. Of the 500 data is divided into 450 data for training data and 50 data for test data. By using the method Longest Common Subsequences for word improvement and the method Support Vector Machine for classification with the best parameter values, namely learning rate (γ) = 0.0001, lambda (λ) = 0.1, complexity (C) = 0.001, epsilon (ϵ ) = 0.0001 and the maximum iteration = 50 can produce the average value of the evaluation results, namely precision = 0.5653, recall = 0.948, f-measure = 0.7047 and   accuracy = 0.598. The test results show that by adding method of Longest Common Subsequences for word improvement, it can increase the level of accuracy which was previously only 0.59 to 0.598.</em></p>

No abstract

Nowadays, there are still often unbalanced nutritional problems such as overnutrition or malnutrition. Many factors can affect it, one of which is an unbalanced diet. One solution that can be done is a system for optimizing nutritional needs. In this study, the method used for optimization is genetic algorithms. Genetic algorithms are one of the metaheuristic methods that are often used for optimization problems. A particular chromosome representation is designed to provide suitable solutions. The system can provide food recommendations with nutrients close to a person's nutritional needs by using the genetic algorithm. Based on the test results obtained, the difference in nutrition from food recommendations with nutritional needs is below 5 percent.

Remote Sensing is one of the relatively complex problems in Machine Learning because of spatial patterns and intricate geometric structures of the data that make semantic understanding meaning essential in the remote sensing community. CNN is one of the Machine Learning methods often used in Remote Sensing problems. However, high-resolution aerial view classification often leverages large-scale data with a huge number of parameters of the CNN model. That large number of parameters makes it hard to be applied to remote imaging peripherals because it requires high capacity of storage and memory. We propose a training framework to solve that problem that produces a model with minimum parameters without sacrificing accuracy. In this study, the three CNN architectures are used to be compared: ResNet, Inception, and EfficientNet. Furthermore, the most efficient CNN backbone then leverages Freezing layers, adding Weighted Loss and Sparse Regularization during training and then Pruning after training. Using this method on the AID Dataset, the best results are achieved by EfficientNet-B0 with Freeze 2 Layer, INS Weighted Loss, Sparse regularization with lambda = 0.001, and Global Unstructured Conv2D Pruning with an accuracy of 95.86% on test data with total parameters of 2,463,501. This study proves that Weighted Loss and Sparse Regularization can help the model to improve accuracy while Pruning enhances efficiency by cutting the model parameters into a few.