出版社：Academy & Industry Research Collaboration Center (AIRCC)
摘要：In this work, the TREPAN algorithm is enhanced and extended for extracting decision treesfrom neural networks. We empirically evaluated the performance of the algorithm on a set ofdatabases from real world events. This benchmark enhancement was achieved by adaptingSingle-test TREPAN and C4.5 decision tree induction algorithms to analyze the datasets. Themodels are then compared with X-TREPAN for comprehensibility and classification accuracy.Furthermore, we validate the experimentations by applying statistical methods. Finally, themodified algorithm is extended to work with multi-class regression problems and the ability tocomprehend generalized feed forward networks is achieved.
