| Title: | A discriminant distance based composite vector selection method for odor classification |
| Address: | "Department of Applied Computer Engineering, Dankook University, 126 Jukjeon-dong, Suji-gu, Yongin-si, Gyeonggi-do 448-701, Korea. choisi@dankook.ac.kr. Electrical Engineering, Kookmin University 2, 861-1, Jeongeung-dong, Songbuk-gu, Seoul 136-702, Korea. gm1004@kookmin.ac.kr" |
| ISSN/ISBN: | 1424-8220 (Electronic) 1424-8220 (Linking) |
| Abstract: | "We present a composite vector selection method for an effective electronic nose system that performs well even in noisy environments. Each composite vector generated from a electronic nose data sample is evaluated by computing the discriminant distance. By quantitatively measuring the amount of discriminative information in each composite vector, composite vectors containing informative variables can be distinguished and the final composite features for odor classification are extracted using the selected composite vectors. Using the only informative composite vectors can be also helpful to extract better composite features instead of using all the generated composite vectors. Experimental results with different volatile organic compound data show that the proposed system has good classification performance even in a noisy environment compared to other methods" |
| Keywords: | Acetone/analysis *Algorithms *Electronic Nose Odorants/*analysis Polymers/analysis Principal Component Analysis Time Factors Volatile Organic Compounds/analysis; |
| Notes: | "MedlineChoi, Sang-Il Jeong, Gu-Min eng Research Support, Non-U.S. Gov't Switzerland 2014/04/22 Sensors (Basel). 2014 Apr 17; 14(4):6938-51. doi: 10.3390/s140406938" |
