VEXAS: VISTA EXtension to Auxiliary Surveys - Data Release 2: Machine-learning based classification of sources in the Southern Hemisphere

Home

All issues

Volume 651 (July 2021)

A&A, 651 (2021) A69

Full HTML

Free Access

Table 4.

Description of the 32 individual classifiers that we combined into the ensemble learning, varying the algorithms, classification problem, input bands, feature sets, the use of imputation, and the auxiliary training sample.

Mod.	Method	Classes	Input bands	Features	Imputation	Aux.
#						sample
Complete models (one for each algorithm)
1	ANN	`STAR, QSO, GAL`	J, K_S, g, r, i, z, y/u^(a), W1, W2	sc.mags, colours, P_STAR	`True`	`True`
2	kNN	`STAR, QSO, GAL`	J, K_S, g, r, i, z, y/u^(a), W1, W2	sc.mags, colours, P_STAR	`True`	`True`
3	CB	`STAR, QSO, GAL`	J, K_S, g, r, i, z, y/u^(a), W1, W2	sc.mags, colours, P_STAR	`True`	`True`

Different band-sets
4	ANN	`STAR, QSO, GAL`	J, K_S, g, r, i, z, y/u^(a)	sc.mags, colours, P_STAR	`True`	`True`
5	kNN	`STAR, QSO, GAL`	J, K_S, g, r, i, z, y/u^(a)	sc.mags, colours, P_STAR	`True`	`True`
6	CB	`STAR, QSO, GAL`	J, K_S, g, r, i, z, y/u^(a)	sc.mags, colours, P_STAR	`True`	`True`
7	ANN	`STAR, QSO, GAL`	g, r, i, z, y/u^(a), W1, W2	sc.mags, colours, P_STAR	`True`	`True`
8	kNN	`STAR, QSO, GAL`	g, r, i, z, y/u^(a), W1, W2	sc.mags, colours, P_STAR	`True`	`True`
9	CB	`STAR, QSO, GAL`	g, r, i, z, y/u^(a), W1, W2	sc.mags, colours, P_STAR	`True`	`True`
10	ANN	`STAR, QSO, GAL`	J, K_s, g, r, i, z, W1, W2	sc.mags, colours, P_STAR	`True`	`True`
11	KNN	`STAR, QSO, GAL`	J, K_s, g, r, i, z, W1, W2	sc.mags, colours, P_STAR	`True`	`True`
12	CB	`STAR, QSO, GAL`	J, K_s, g, r, i, z, W1, W2	sc.mags, colours, P_STAR	`True`	`True`

Different classification problems
13	ANN	`STAR, QSO+GAL`	J, K_S, g, r, i, z, y/u^(a), W1, W2	sc.mags, colours, P_STAR	`True`	`True`
14	kNN	`STAR, QSO+GAL`	J, K_S, g, r, i, z, y/u^(a), W1, W2	sc.mags, colours, P_STAR	`True`	`True`
15	CB	`STAR, QSO+GAL`	J, K_S, g, r, i, z, y/u^(a), W1, W2	sc.mags, colours, P_STAR	`True`	`True`
16	ANN	`STAR+GAL, QSO`	J, K_S, g, r, i, z, y/u^(a), W1, W2	sc.mags, colours, P_STAR	`True`	`True`
17	kNN	`STAR+GAL, QSO`	J, K_S, g, r, i, z, y/u^(a), W1, W2	sc.mags, colours, P_STAR	`True`	`True`
18	CB	`STAR+GAL, QSO`	J, K_S, g, r, i, z, y/u^(a), W1, W2	sc.mags, colours, P_STAR	`True`	`True`
19	ANN	`STAR+QSO,GAL`	J, K_S, g, r, i, z, y/u^(a), W1, W2	sc.mags, colours, P_STAR	`True`	`True`
20	kNN	`STAR+QSO,GAL`	J, K_S, g, r, i, z, y/u^(a), W1, W2	sc.mags, colours, P_STAR	`True`	`True`
21	CB	`STAR+QSO, GAL`	J, K_S, g, r, i, z, y/u^(a), W1, W2	sc.mags, colours, P_STAR	`True`	`True`

The effect of imputation and auxiliary sample
22	ANN	`STAR, QSO, GAL`	J, K_S, g, r, i, z, y/u^(a), W1, W2	sc.mags, colours, P_STAR	`True`	`False`
23	kNN	`STAR, QSO, GAL`	J, K_S, g, r, i, z, y/u^(a), W1, W2	sc.mags, colours, P_STAR	`True`	`False`
24	CB	`STAR, QSO, GAL`	J, K_S, g, r, i, z, y/u^(a), W1, W2	sc.mags, colours, P_STAR	`True`	`False`
25	CB	`STAR, QSO, GAL`	J, K_S, g, r, i, z, y/u^(a), W1, W2	sc.mags, colours, P_STAR	`False`	`True`
26	CB	`STAR, QSO, GAL`	J, K_S, g, r, i, z, y/u^(a), W1, W2	sc.mags, colours, P_STAR	`False`	`False`

Classification without stellarity
27	ANN	`STAR, QSO, GAL`	J, K_S, g, r, i, z, y/u^(a), W1, W2	sc.mags, colours	`True`	`True`
28	kNN	`STAR, QSO, GAL`	J, K_S, g, r, i, z, y/u^(a), W1, W2	sc.mags, colours	`True`	`True`
29	CB	`STAR, QSO, GAL`	J, K_S, g, r, i, z, y/u^(a), W1, W2	sc.mags, colours	`True`	`True`

Classification without magnitudes
30	ANN	`STAR, QSO, GAL`	J, K_S, g, r, i, z, y/u^(a), W1, W2	colours, P_STAR	`True`	`True`
31	kNN	`STAR, QSO, GAL`	J, K_S, g, r, i, z, y/u^(a), W1, W2	colours, P_STAR	`True`	`True`
32	CB	`STAR, QSO, GAL`	J, K_S, g, r, i, z, y/u^(a), W1, W2	colours, P_STAR	`True`	`True`

Notes.

^(a)

We use the y band for PS and DES, and the u band for SM.

Current usage metrics show cumulative count of Article Views (full-text article views including HTML views, PDF and ePub downloads, according to the available data) and Abstracts Views on Vision4Press platform.

Data correspond to usage on the plateform after 2015. The current usage metrics is available 48-96 hours after online publication and is updated daily on week days.

Initial download of the metrics may take a while.