Cost-sensitive Greedy Network-growing Algorithm and Its Application to Economic Data Analysis

R. Kamimura and S. Hashimoto (Japan)

Keywords

greedy networkgrowing, mutual information maximization, competitive learning, cost, economicdata

Abstract

In this paper, we propose a new type of greedy network growing algorithm. In this new model, a network grows by absorbing as much information as possible. In addi tion, the associate cost needed to maximize information is controlled to be minimized. We applied the method to an economical data analysis. Experimental results showed that the method can explicitly classify economic data into several groups. The groups obtained by this method could clearly be interpreted.

Important Links:



Go Back