Garuda - Garba Rujukan Digital

ADAPTIVE ANT COLONY OPTIMIZATION BASED GRADIENT FOR EDGE DETECTION Febri Liantoni; Kartika Candra Kirana; Tri Hadiah Muliawati
Jurnal Ilmu Komputer dan Informasi Vol 7, No 2 (2014): Jurnal Ilmu Komputer dan Informasi (Journal of Computer Science and Information)
Publisher : Faculty of Computer Science - Universitas Indonesia

Ant Colony Optimization (ACO) is a nature-inspired optimization algorithm which is motivated by ants foraging behavior. Due to its favorable advantages, ACO has been widely used to solve several NP-hard problems, including edge detection. Since ACO initially distributes ants at random, it may cause imbalance ant distribution which later affects path discovery process. In this paper an adaptive ACO is proposed to optimize edge detection by adaptively distributing ant according to gradient analysis. Ants are adaptively distributed according to gradient ratio of each image regions. Region which has bigger gradient ratio, will have bigger number of ant distribution. Experiments are conducted using images from various datasets. Precision and recall are used to quantitatively evaluate performance of the proposed algorithm. Precision and recall of adaptive ACO reaches 76.98 % and 96.8 %. Whereas highest precision and recall for standard ACO are 69.74 % and 74.85 %. Experimental results show that the adaptive ACO outperforms standard ACO which randomly distributes ants.

Modifikasi Ant Colony Optimization Berdasarkan Gradient Untuk Deteksi Tepi Citra Febri Liantoni; Nanik Suciati; Chastine Fatichah
Jurnal Buana Informatika Vol. 6 No. 3 (2015): Jurnal Buana Informatika Volume 6 Nomor 3 Juli 2015
Publisher : Universitas Atma Jaya Yogyakarta

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.24002/jbi.v6i3.435

Abstract. Ant Colony Optimization (ACO) is an optimization algorithm which can be used for image edge detection. In traditional ACO, the initial ant are randomly distributed. This condition can cause an imbalance ants distribution. Based on this problem, a modified ant distribution in ACO is proposed to optimize the deployment of ant based gradient. Gradient value is used to determine the placement of the ants. Ants are not distributed randomly, but are placed in the highest gradient. This method is expected to be used to optimize the path discovery. Based on the test results, the use of the proposed ACO modification can obtain an average value of the Peak Signal to Noise Ratio (PSNR) of 12.724. Meanwhile, the use of the traditional ACO can obtain an average value of PSNR of 12.268. These results indicate that the ACO modification is capable of generating output image better than traditional ACO in which ants are initially distributed randomly.Keywords: Ant Colony Optimization, gradient, Edge Detection, Peak Signal to Noise RatioÂ Abstrak. Ant Colony Optimization (ACO) merupakan algoritma optimasi, yang dapat digunakan untuk deteksi tepi pada citra Pada ACO tradisional, semut awal disebarkan secara acak. Kondisi ini dapat menyebabkan ketidakseimbangan distribusi semut. Berdasarkan permasalahan tersebut, modifikasi distribusi semut pada ACO diusulkan untuk mengoptimalkan penempatan semut berdasarkan gradient. Nilai gradient digunakan untuk menentukan penempatan semut. Semut tidak disebar secara acak akan tetapi ditempatkan di gradient tertinggi. Cara ini diharapkan dapat digunakan untuk optimasi penemuan jalur. Berdasarkan hasil uji coba, dengan menggunakan ACO modifikasi yang diusulkan dapat diperoleh nilai rata-rata Peak Signal to Noise Ratio (PSNR) 12,724. Sedangkan, menggunakan ACO tradisional diperoleh nilai rata-rata PSNR 12,268. Hasil ini menunjukkan bahwa ACO modifikasi mampu menghasilkan citra keluaran yang lebih baik dibandingkan ACO tradisional yang sebaran semut awalnya dilakukan secara acak.Kata Kunci: Ant Colony Optimization, gradient, deteksi tepi, Peak Signal to Noise Ratio

Pengembangan Metode Ant Colony Optimization Pada Klasifikasi Tanaman Mangga Menggunakan K-Nearest Neighbor Febri Liantoni; Luky Agus Hermanto
Jurnal Buana Informatika Vol. 8 No. 4 (2017): Jurnal Buana Informatika Volume 8 Nomor 4 Oktober 2017
Publisher : Universitas Atma Jaya Yogyakarta

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.24002/jbi.v8i4.1443

Abstract. Leaf is one important part of a plant normally used to classify the types of plants. The introduction process of mango leaves of mangung and manalagi mango is done based on the leaf edge image detection. In this research the conventional edge detection process was replaced by ant colony optimization method. It is aimed to optimize the result of edge detection of mango leaf midrib and veins image. The application of ant colony optimization method successfully optimizes the result of edge detection of a mango leaf midrib and veins structure. This is demonstrated by the detection of bony edges of the leaf structure which is thicker and more detailed than using a conventional edge detection. Classification testing using k-nearest neighbor method obtained 66.67% accuracy.Keywords: edge detection, ant colony optimization, classification, k-nearest neighbor.Abstrak. Pengembangan Metode Ant Colony Optimization Pada Klasifikasi Tanaman Mangga Menggunakan K-Nearest Neighbor. Daun merupakan salah satu bagian penting dari tanaman yang biasanya digunakan untuk proses klasifikasi jenis tanaman. Proses pengenalan daun mangga gadung dan mangga manalagi dilakukan berdasarkan deteksi tepi citra struktur tulang daun. Pada penelitian ini proses deteksi tepi konvensional digantikan dengan metode ant colony optimization. Hal ini bertujuan untuk optimasi hasil deteksi tepi citra tulang daun mangga. Penerapan metode ant colony optimization berhasil mengoptimalkan hasil deteksi tepi struktur tulang daun mangga. Hal ini ditunjukkan berdasarkan dari hasil deteksi tepi citra struktur tulang daun yang lebih tebal dan lebih detail dibandingkan menggunakan deteksi tepi konvensional. Pengujian klasifikasi dengan metode k-nearest neighbor didapatkan nilai akurasi sebesar 66,67%.Kata Kunci: deteksi tepi, ant colony optimization, klasifikasi, k-nearest neighbor.

KLASIFIKASI DAUN HERBAL MENGGUNAKAN METODE NAÏVE BAYES CLASSIFIER DAN KNEAREST NEIGHBOR Febri Liantoni; Hendro Nugroho
Jurnal Simantec Vol 5, No 1 (2015)
Publisher : Universitas Trunojoyo Madura

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.21107/simantec.v5i1.1009

ABSTRAKPerkembangan ilmu tanaman telah mengalami kemajuan yang pesat, khususnya ilmu mengenai tanaman herbal. Tanaman herbal memiliki banyak manfaat bagi kehidupan manusia yaitu sebagai penyedian oksigen, bahan makanan, obat-obatan, maupun bahan kosmetika. Untuk mengetahui jenis-jenis tanaman herbal dapat dilakukan dengan proses klasifikasi. Klasifikasi tanaman herbal dapat dilakukan dengan cara mengidentifikasi bentuk citra daun dari tanaman tersebut. Proses klasifikasi tanaman herbal memerlukan ekstraksi fitur dari bentuk dari tanaman tersebut. Pada penelitian ini, fitur invariant moment dan fitur geometri digunakan untuk ekstraksi fitur daun herbal. Fitur yang digunakan berjumlah 10 fitur. Ada beberapa macam metode klasifikasi yang biasa digunakan. Pada penelitian ini metode klasifikasi yang digunakan adalah metode Naïve Bayes Classifier dan K-Nearest Neighbor (KNN). Metode Naïve Bayes Classifier merupakan metode Bayesian Learning yang paling cepat dan sederhana. Sedangkan metode KNN dapat melakukan klasifikasi dengan cepat berdasarkan jarak terdekat diantara objek data. Berdasarkan hasil uji coba yang dilakukan, penggunaan metode Naïve Bayes Classifier didapatkan nilai akurasi sebesar 75%, sedangkan dengan menggunakan metode K-Nearest Neighbor didapatkan nilai akurasi sebesar 70,83%. Hal ini menunjukkan bahwa kinerja metode Naïve Bayes Classifier lebih baik dibandingkan metode KNN.Kata Kunci: Fitur Invariant Moment, Fitur Geometri, Naïve Bayes Classifier, K-Nearest Neighbor, Bayesian Learning. ABSTRACTScience of the plant has made progress, particularly knowledge about herbs. Herb has many benefits for human life as provision of oxygen, foodstuffs, pharmaceuticals, and cosmetics. To determine the types of herbs with the classification process. Classification of herbs conducted by identifying the shape of the image of the leaves of these plants. Herbal plant classification process requires the extraction of features from the shape of plant. In this study, moment invariant features and geometrical feature is used for feature extraction of herbal leaves.Features used amounted to 10 features. There are several kinds of commonly used classification method. In this study, the classification method used is the method Naïve Bayes classifier and K-Nearest Neighbor (KNN). Naïve Bayes classifier is Bayesian Learning method of the most rapid and simple. While the KNN method can perform fast classification is based on the shortest distance between data objects. Based on the results of tests conducted, the use of methods Naïve Bayes Classifier accuracy values obtained by 75%, while using K-Nearest Neighbor value obtained an accuracy of 70.83%.These results indicate that the performance of Naïve Bayes Classifier method is better than KNN method.Keywords: Invariant Moment Feature, Geometrical Feature, Naïve Bayes Classifier, K-Nearest Neighbor, Bayesian Learning

Klasifikasi Daun Dengan Perbaikan Fitur Citra Menggunakan Metode K-Nearest Neighbor Febri Liantoni
Ultimatics : Jurnal Teknik Informatika Vol 7 No 2 (2015): Ultimatics: Jurnal Ilmu Teknik Informatika
Publisher : Faculty of Engineering and Informatics, Universitas Multimedia Nusantara

Plants are the most important part in life on earth as oxygen supplier to breathe, groceries, fuel, medicine and more. Plants can be classified based on its leaves shape. Classification process is required well data extraction feature, so it needs fixing feature process at pre-processing level. Combining median filter and image erosion is used for fixing feature process. Whereas for feature extraction is used invariant moment method. In this research, it is used leaves classification based on leaves edge shape. K-Nearest Neighbor Method (KNN) is used for leaves classification process. KNN method is chosen because this method is known rapid in training data, effective for large training data, simple and easy to learn. Testing the result of leaves classification from image which is on dataset has been built to get accuracy value about 86,67%. Index Terms—Classification, Median Filter, Invariant Moment, K-Nearest Neighbor.

Perbandingan Metode Single Exponential Smoothing dan Metode Holt untuk Prediksi Kasus COVID-19 di Indonesia Nur Hijrah As Salam Al Ihsan; Hanifah Hanun Dzakiyah; Febri Liantoni
Ultimatics : Jurnal Teknik Informatika Vol 12 No 2 (2020): Ultimatics : Jurnal Teknik Informatika
Publisher : Faculty of Engineering and Informatics, Universitas Multimedia Nusantara

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.31937/ti.v12i2.1689

Coronavirus disease (COVID-19) was first discovered in December 2019 in Wuhan, China, and spread so quickly into a pandemic. This outbreak has spread to 24 other countries, including Indonesia. Its spread is very fast, so a co-19 prediction study is needed to be able to make the right policy. To be able to predict the number of COVID-19 cases can be done with the Forecasting Technique. The purpose of this study is to forecast and compare Single Exponential Smoothing and Double Exponential Smoothing ¬ against the number of COVID-19 cases in Indonesia. The results of this study can be used as consideration for policymaking in dealing with the spread of COVID-19. Distribution predictions are based on data released by the Indonesian National Disaster Management Agency (BNPB) in the first 100 days of COVID-19 deployment. The results of this study are the Double Exponential Smoothing method is more accurate than the Single Exponential Smoothing method because the forecasting results show an increase from the previous data. And the percentage of errors (MAPE) obtained is significantly smaller.

Analisis Clustering Pengelompokan Penjualan Paket Data Menggunakan Metode K-Means Dimas Galang Ramadhan; Indri Prihatini; Febri Liantoni
Ultimatics : Jurnal Teknik Informatika Vol 13 No 1 (2021): Ultimatics : Jurnal Teknik Informatika
Publisher : Faculty of Engineering and Informatics, Universitas Multimedia Nusantara

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.31937/ti.v13i1.1981

At present with the COVID-19 pandemic situation that requires all activities based in the network, starting from work, college, school, everything is based on the network. Certain provider users will experience excessive data plan usage. This also has an effect on a counter that sells data packages, which must provide several data package services in accordance with current conditions. This research was conducted to analyze the grouping of sales of data packages that are often purchased by customers in a counter by using the K-Means method. The K-Means method is used because the K-Means algorithm is not influenced by the order of the objects used, this is proven when the writer tries to determine the initial cluster center randomly from one of the objects in the first calculation. sales of data packages at a counter. Variables used include Price, Active period, and number of data packages. The K-Means Cluster Analysis algorithm is basically applied to the problem of understanding consumer needs, identifying the types of data package products that are often purchased. The K-Means algorithm can be used to describe the characteristics of each group by summarizing a large number of objects so that it is easier.

Title

Found 7 Documents
Search

Abstract

Abstract

Abstract

Abstract

Abstract

Abstract

Abstract

Title Search

Found 7 Documents Search

Abstract

Abstract

Abstract

Abstract

Abstract

Abstract

Abstract

Title

Found 7 Documents
Search