修正型自組特徵映射神經網路應用於離散數據的聚類

The Modified Self-Organizing Map Neural Network Applied To The Discrete Data Clustering

程家信(Chia-Hsin Cheng)

指導教授 ： 紀美秀

國立中正大學/理學院/數學系研究所/碩士(2012年)

摘要

本文的目的在於解決新型自組特徵映射網路的缺點，例如:

1. 資料與類別的性質差異大。
2. 在分類時有些資料沒有歸屬。

因此，本文提出一個方法刪除沒有使用到的類別，並且新增類別將未分類的資料進行分類。這樣的方法不僅能使資料與類別的性質差異縮小，而且能使所有資料皆有歸屬。

關鍵字

類神經網路

並列摘要

The purpose of this paper is to solve the shortcomings of the novel self-organizing map neural network. For example:

1. Data and category are in large differences in nature.
2. Some of the data are without attribution .

Therefore, this paper presents a method to remove unused categories and to add new categories for those un-attribution data . The method not only narrows the differences in the nature between data and categories , but also enables all data attribution .

並列關鍵字

artificial neural network

參考文獻

4 M.H. Ghaseminezhad, A. Karami, A novel self-organizing map (SOM) neural network for discrete groups of data clustering,Applied Soft Computing 11 (2011) 3771–3778

5 T. Furukawa, SOM of SOMs: self-organizing map which maps a group of self organizing maps, in: Proc. International Conference on Artificial Neural Networks,2005, pp. 391–396.

6 Anders Krogh , What are artificial neural networks? NATURE BIOTECHNOLOGY VOLUME 26 NUMBER 2 FEBRUARY 2008 195-197

1 葉怡成，2009，類神經網路模式應用與實作，儒林圖書有限公司

2 張斐章、張麗秋，2010，類神經網路導論:原理與應用，滄海書局