Automated quasar continuum estimation using neural networks - A comparative study of deep-learning architectures

Open Access

Table D.1

Architecture of the optimized autoencoders applied to WEAVE (red and full spectra) and VIPERS data.

Model	Type	Size	Additional Info
WEAVE	Input	3920
red part	Masking	3920	Mask value: 0
	Dense	512	Act.: ReLU
	Dense	128	Act.: ReLU
	Dense	1024	Act.: ELU
	Dense	4900
WEAVE	Input	4900
full spectra	Masking	4900	Mask value: 0
	Dense	512	Act.: ReLU
	Dense	128	Act.: ReLU
	Dense	1024	Act.: ELU
	Dense	4900
VIPERS	Input	2000
	Masking	2000	Mask value: 0
	Dense	896	Act.: ReLU
	Dense	1024	Act.: ReLU
	Dense	512	Act.: ReLU
	Dense	64	Act.: ELU
	Dense	640	Act.: ReLU
	Dense	512	Act.: ReLU
	Dense	768	Act.: ReLU
	Dense	2000

Notes. The activation functions (Act. value in the table) used are the exponential linear unit (ELU) and the rectified linear unit (ReLU) function.

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.