Representation learning for automated spectroscopic redshift estimation

Open Access

Table 1.

Success rate of the redshift estimation algorithms in the different S/N cases, where failure is defined as z_est − z_true > 0.003 (1 + z_true).

S/N = 20				S/N = 5				S/N = 2
Total DF success: 19479 (97.4%)				Total DF success: 13125 (65.6%)				Total DF success: 5586 (27.9%)
Total DL success: 18668 (93.3%)				Total DL success: 18694 (93.5%)				Total DL success: 17822 (89.1%)
Total DAE success: 19954 (99.8%)				Total DAE success: 19880 (99.4%)				Total DAE success: 15545 (77.7%)

Total “best” success: 19945 (99.7%)				Total “best” success: 19895 (99.5%)				Total “best” success: 18630 (93.1%)

DF	DL	DAE		DF	DL	DAE		DF	DL	DAE

✓	✓	✓	18185	✓	✓	✓	12238	✓	✓	✓	5203
✓	✓	×	14	✓	✓	×	18	✓	✓	×	87
✓	×	✓	1264	✓	×	✓	856	✓	×	✓	256
×	✓	✓	455	×	✓	✓	6357	×	✓	✓	9082
✓	×	×	16	✓	×	×	13	✓	×	×	40
×	✓	×	14	×	✓	×	81	×	✓	×	3450
×	×	✓	50	×	×	✓	429	×	×	✓	1004
×	×	×	2	×	×	×	8	×	×	×	878

Notes. The “best” algorithm combines DL/DAE results as described in Sect. 6.5 and shown in Fig. 10. The lower half of the table shows the partial statistics for different combinations of algorithm success; for example, the first line counts the number of galaxies for which the three algorithms succeeded in retrieving the true redshift, the second line counts the number of galaxies where DF and DL succeeded while DAE failed, and so forth.

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.