Spectral Selection Method Based on Ant Colony-Genetic Algorithm

Qing HUANG; He-ru XUE; Jiang-ping LIU; Mei-chen LIU; Peng-wei HU; De-gang SUN

doi:10.3964/j.issn.1000-0593(2022)07-2262-07

Journals >Spectroscopy and Spectral Analysis >Volume 42 >Issue 7 >Page 2262 > Article

Spectroscopy and Spectral Analysis
Vol. 42, Issue 7, 2262 (2022)

Spectral Selection Method Based on Ant Colony-Genetic Algorithm

Qing HUANG^1、1;, He-ru XUE^{1、1; *;}, Jiang-ping LIU^{1、1; *;}, Mei-chen LIU^1、1;, Peng-wei HU^1、1;, and De-gang SUN^2、2;

Author Affiliations

¹1. College of Computer and Information Engineering, Inner Mongolia Agricultural University, Huhhot 010000, China

²2. College of Information Engineering, Shangdong HuaYu University of Technology, Dezhou 253000, China

show less

DOI: 10.3964/j.issn.1000-0593(2022)07-2262-07 Cite this Article

Qing HUANG, He-ru XUE, Jiang-ping LIU, Mei-chen LIU, Peng-wei HU, De-gang SUN. Spectral Selection Method Based on Ant Colony-Genetic Algorithm[J]. Spectroscopy and Spectral Analysis, 2022, 42(7): 2262 Copy Citation Text

show less

Fig. 1. Original spectra of samples of six different brands
(a): Yilizhennong; (b): Yiliskim milk; (c): Mengniu high calcium milk; (d): QQ star; (e): The land; (f): Telunsu

Download full size | View in the Article

Fig. 2. PLS-ACO-GA algorithm flowchart

Download full size | View in the Article

Fig. 3. PLS-ACO band contributionrate

Download full size | View in the Article

Fig. 4. PLS-ACO-GA band contributionrate

Download full size | View in the Article

Fig. 5. MSC-(PLS-ACO-GA)-RF prediction results

Download full size | View in the Article

预处理方法	$R_{c}^{2}$	$R_{p}^{2}$	RMSECV	RMSEP
Raw data	0.999 1	0.987 3	0.046 6	0.143 7
1^st Der	0.998 3	0.973 6	0.064 7	0.207 4
2^st Der	0.999 1	0.954 6	0.046 7	0.270 0
S-G	0.998 8	0.988 8	0.053 4	0.135 1
MSC	0.999 9	0.993 4	0.015 5	0.103 8
SNV	0.999 0	0.988 0	0.049 4	0.139 6

Table 1. The prediction results of the PLS regression model for the milk fat content of different pretreatment methods

View in the Article

max-features	波长数	$R_{p}^{2}$	RMSEP	主成分数
Full	125	0.993 4	0.103 8	24
10	21	0.996 2	0.077 8	10
20	18	0.997 0	0.069 2	16
30	41	0.997 8	0.058 5	12

Table 2. PLS regression model results with PLS-ACO and max-features set to 10, 20 and 30 respectively

View in the Article

max-features	波长数	$R_{p}^{2}$	RMSEP	主成分数
Full	125	0.934 4	0.103 8	24
10	16	0.996 8	0.075 2	8
20	16	0.997 6	0.062 2	14
30	32	0.998 0	0.056 5	12

Table 3. PLS regression model results with PLS-ACO-GA and max-features set to 10, 20 and 30 respectively

View in the Article

方法	波长数	$R_{p}^{2}$	RMSEP	主成分数
Full	125	0.934 4	0.103 8	24
CARS	40	0.995 1	0.088 7	12
GA	43	0.997 1	0.068 5	14
ACO	22	0.996 2	0.072 8	12
PLS-ACO	18	0.997 0	0.077 8	16
PLS-ACO-GA	16	0.997 6	0.062 2	14

Table 4. PLS regression model result sunder different feature selection methods

View in the Article

方法	波长数	$R_{p}^{2}$	RMSEP
Full-MLR	125	0.991 4	0.118 0
PLS-ACO-MLR	18	0.997 0	0.069 2
PLS-ACO-GA-MLR	16	0.997 6	0.062 3

Table 5. PLS-ACO and PLS-ACO-GA MLR results

View in the Article

方法	波长数	$R_{p}^{2}$	RMSEP
Full-RFR	125	0.999 963	0.007 6
PLS-ACO-RFR	18	0.999 974	0.006 4
PLS-ACO-GA-RFR	16	0.999 994	0.003 0

Table 6. PLS-ACO and PLS-ACO-GA RFR results

Download Citation

Save the article for my favorites

Paper Information