Claim Missing Document
Check
Articles

Found 2 Documents
Search
Journal : Jurnal Ilmu Komputer

PENERAPAN MULTI LAYER PERCEPTRON DALAM ANOTASI IMAGE SECARA OTOMATIS Agus Muliantara; I Made Widiartha
Jurnal Ilmu Komputer Vol. 4, No. 2 September 2011
Publisher : Informatics Department, Faculty of Mathematics and Natural Sciences, Udayana University

Show Abstract | Download Original | Original Source | Check in Google Scholar | Full PDF (351.362 KB)

Abstract

Penentuan anotasi terhadap suatu image (image annotation) merupakan topik penelitian yang berkembang pesat akhir-akhir ini. Permasalahan yang ada dalam menentukan anotasi sebuah image adalah dalam hal penentuan fitur dan metode yang digunakan agar hasil anotasi yang didapat sesuai dengan yang diharapkan oleh pengguna.Dalam penelitian ini akan diimplementasikan suatu model untuk memprediksi anotasi suatu image. Penentuan fitur suatu image dilakukan dengan menggunakan metode color quantization dan multi-level wavelet transform. Dalam melakukan prediksi anotasi suatu image, dilakukan dengan mengimplementasikan metode Multi Layer Perceptron (MLP).Untuk mengevaluasi performance dari model yang diimplementasikan digunakan data image sebanyak 453. Hasil penelitian yang telah dilakukan menunjukkan bahwa tingkat akurasi untuk prediksi anotasi oleh MLP adalah sebesar 81%.
PENERAPAN METODE ANT COLONY OPTIMZATION PADA METODE K-HARMONIC MEANS UNTUK KLASTERISASI DATA I Made Kunta Wicaksana; I Made Widiartha
Jurnal Ilmu Komputer Vol. 5, No. 1 April 2012
Publisher : Informatics Department, Faculty of Mathematics and Natural Sciences, Udayana University

Show Abstract | Download Original | Original Source | Check in Google Scholar | Full PDF (308.332 KB)

Abstract

Data can be classified into several clusters, better known as Data Clustering using several methods, one of which is referred to as K-Means method (KM). It is one of the popular data clustering method. Its implementation is simple and can cope with a great number of data and the process is relatively short. However, KM has several weaknesses; the clustering result is sensitive to the initialization of the cluster center and leads to optimal local. It is the betterment of KM method referred to as K-Harmonic Means (KHM). Although it can minimize in the initialization, it could not overcome the problem of optimal local yet.Ant Colony Optimization (ACO) is an ant algorithm used to form a colony. ACO could avoid the problem of local optimal and was proved to have global solution. In this study, an algorithm was applied to clusterizing the ACO and KHM-based data referred to as ACOKHM. The performance of ACOKHM was compared to the algorithms of ACO and KHM using five data sets. The ACOKHM algorithm was proved to have better performance than ACO and KHM, in which ACOKHM could maximize the cluster center which directs to optimal global.