Photometric redshift estimation via deep learning - Generalized and pre-classification-less, image based, fully probabilistic redshifts

Free Access

Table 4

Results of Experiment 2 (quasars).

Criterion	Model	Bias(Δz)	σ²(Δz)	Mad(Δz)	rmse(Δz)	Bias(Δz_norm)	σ²(Δz_norm)	Mad(Δz_norm)	rmse(Δz_norm)

	RF	0.007	0.217	0.145	0.466	–0.033	0.048	0.050	0.222
Mean	MDN	–0.002	0.216	0.156	0.465	–0.037	0.048	0.054	0.223
	DCMDN	0.011	0.168	0.128	0.411	–0.023	0.035	0.045	0.189

	RF	0.002	0.319	0.087	0.564	–0.026	0.066	0.031	0.258
First mode	MDN	–0.058	0.282	0.095	0.535	–0.052	0.068	0.034	0.267
	DCMDN	–0.043	0.206	0.095	0.456	–0.038	0.048	0.034	0.222

	RF	0.005	0.162	0.111	0.402	–0.024	0.037	0.039	0.194
Selected	MDN	–0.010	0.098	0.086	0.314	–0.017	0.021	0.030	0.145
	DCMDN	0.004	0.047	0.075	0.217	–0.004	0.009	0.026	0.095

Notes. Similarly to experiment 1, the traditional scores are presented. The score have been calculated with the mean, the most dominant mode and the mean of the selected unambiguous objects.

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.