@article{oai:miyazaki-u.repo.nii.ac.jp:00002653,
 author = {山森, 一人 and Yamamori, Kunihito and 吉田, 卓矢 and 相川, 勝 and Aikawa, Masaru and Yoshida, Takuya},
 journal = {宮崎大学工学部紀要, Memoirs of Faculty of Engineering, University of Miyazaki},
 month = {Aug},
 note = {Recently, the number of digital images we have is increasing as popularizing cellular phone with digital camera. To manage these digital images, some researches investigate new technologies for automatic image classification. Since image classification technologies depends on target images, it is difficult to decide what kinds of techniques and classification threshold should be used. To solve this problem, a technique, called as Genetic Image Network for Image Classification (GIN-IC), has been proposed that builds an image classification algorithm automatically. However, GIN-IC requires a huge computation time. In this paper, we propose a high-speed GIN-IC with master-worker parallel computation model. A master broadcasts an individual to all workers, them workers compute the fitness in parallel by using a part of training images. These fitnesses are collected to the master, and the master calculates final fitness by summing up all the collected fitness. We implement our method in a cluster computer, and evaluate computation time until building suitable algorithm that classifies digital images into two classes. Our method achieved four times faster than that of serial model with a master and six workers.},
 pages = {245--249},
 title = {クラスタ型コンピュータを用いた Genetic Image Network for Image Classification の並列処理による高速化},
 volume = {42},
 year = {2013},
 yomi = {ヤマモリ, クニヒト and ヨシダ, タクヤ and アイカワ, マサル}
}