作者: Mohammed Ramdani , Mohamed Kissi
DOI:
关键词: Data mining 、 Machine learning 、 Sandalwood 、 Artificial intelligence 、 Set (abstract data type) 、 Decision tree 、 Mathematics
摘要: In this work, a new approach to the recognition of odors, based on decision trees, is presented. This has been implemented using database sandalwood molecules and an expert knowledge for these molecules. Decision trees are used learn from powerful rule set determining presence or absence odor. For better prediction, method uses aggregation between molecule descriptors. We apply three operators: Zadeh, Lukasiewicz, Ordered Weighted Averaging. The main olfactory descriptor was correctly predicted by 88 percent elements testing set.