Fuzzy Decision Trees to Extract Features of Odorous Molecules

Abstract : In this paper, an application of fuzzy decision trees to the recognition of odors is presented. Fuzzy decision trees are used to extract and highlight associations in the chemical structure of the smelling and the quality of their odors. Experiments have been conducted with the system Salammbô on data related to the description of molecules, modelized by means of the GESDEM (GEneration and Selection of Descriptor and pattern Elaboration Method). Beside the discontinuous descriptors generated by GESDEM method, we have used another type of descriptors which are continuous: the dimensions of each molecule and theirs ratios as geometric descriptors and logP, Molecular refraction and density as physico-chemical descriptors. These descriptors were measured on the most stable conformation of the molecules.
Document type :
Book sections
Complete list of metadatas

https://hal.archives-ouvertes.fr/hal-01573392
Contributor : Lip6 Publications <>
Submitted on : Wednesday, August 9, 2017 - 2:04:09 PM
Last modification on : Thursday, March 21, 2019 - 12:59:51 PM

Identifiers

Citation

Christophe Marsala, Mohammed Ramdani, Driss Zakarya, Mustapha Tollabi. Fuzzy Decision Trees to Extract Features of Odorous Molecules. Uncertainty in Intelligent and Information Systems, World Scientific, pp.235-249, 2000, ⟨10.1142/9789812792563_0019⟩. ⟨hal-01573392⟩

Share

Metrics

Record views

70