Photometric redshift estimation with convolutional neural networks and galaxy images: Case study of resolving biases in data-driven methods

Open Access

Table 2

Hyperparameters in training.

	Learning rate (*)	# Iterations	Mini-batch size	Balancing threshold
Baseline and step 1 (SDSS)	10⁻⁴	120000	128	–
Baseline and Step 1 (CFHTLS)	10⁻⁴	120000	32	–
Step 2 and Step 3 (SDSS)	2 × 10⁻⁵	40000	32	200
Step 2 and Step 3 (CFHTLS-DEEP)	2 × 10⁻⁵	40000	32	5
Step 2 and Step 3 (CFHTLS-WIDE)	2 × 10⁻⁵	40000	32	100

Notes. ^(*)Reduced to 2 × 10-⁵ after 60 000 iterations for the baseline and step 1.

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.