ILA-3: An Inductive Learning Algorithm with a New Feature Selection Approach

Authors: ,

Abstract: ILA is an inductive learning algorithm proved itself as a powerful algorithm in inductive learning community. It generates the less number of simplest rules compared with other similar algorithms with 100% accuracy; i.e. the generated rules cover all examples in the dataset. But, it becomes inefficient for large datasets. In this paper, a new generation of ILA, called ILA-3, has been developed and tailored with a new feature selection algorithm. This approach takes into consideration the way ILA works and excludes the most irrelevant features from the dataset under consideration while ILA is running, which yields to smaller datasets. Experiments show that significant efficiency improvements (that reached 30% on average) have been gained with ILA-3 over the original ILA with keeping accuracy with acceptable levels.

WSEAS Transactions on Systems and Control, ISSN / E-ISSN: 1991-8763 / 2224-2856, Volume 13, 2018, Art. #21

PDF
XML
Cite
×
Citation Tools
WSEAS AMA
Saleh M. Abu-Soud, Sufyan Almajali, "ILA-3: An Inductive Learning Algorithm with a New Feature Selection Approach," WSEAS Transactions on Systems and Control, vol. 13, pp. 171-185, 2018, DOI:
Saleh M. Abu-Soud, Sufyan Almajali. ILA-3: An Inductive Learning Algorithm with a New Feature Selection Approach. WSEAS Transactions on Systems and Control. 2018;13:171-185.
Copy to Clipboard
Citation copied to Clipboard

Certification

---