Abstract—Nowadays, the method of data mining, widely known as KDD (Knowledge-discovery in databases), is researched actively due to increasing amount of data. Decision tree, which is human-readable and understandable, is widely used in the field of rule extraction, data mining, and pattern recognition. Although decision tree is easy for quickly understanding the data, it has many faults in algorithm itself. In this paper, to compensate for these faults, we combined decision table (rough set theory) and ANN (Artificial Neural Network). With this combined algorithm, we could easily obtain more precise rules from iris data. Our experiment was in this order: 1) Find weights by using back propagation (bp) algorithm. (ANN). 2) Use dimensionality reduction method to delete huge amount of irrelevant data. (Rough set theory and decision table). 3) Make decision tree with pruned weights and rules. With ANN algorithm and decision table (dimensionality reduction method), we could easily notice the most influential (high entropy) character of iris data. With this algorithm, we could extract more precise and pruned rule from iris data sets.
Index Terms—Decision tree, rough set, dimensionality reduction method, decision table, ANN (Artificial Neural Networks), Iris data.
The authors are with the Hankuk Academy of Foreign Studies Wangsan-ri, Mohyon-myeon Cheoinn-gu, Yongin-si, Gyeonggi-do, Korea (e-mail: email@example.com, firstname.lastname@example.org).
Cite: Taehwan Kim and Taeseon Yoon, "Artificial Neural Network Hybrid Algorithm Combimed with Decision Tree and Table," International Journal of Machine Learning and Computing vol.5, no. 6, pp. 471-475, 2015.