Stellar parameter prediction and spectral simulation using machine learning - A systematic comparison of methods with HARPS observational data

Open Access

Table B.2

Decoder architecture from Sedaghat et al. (2021).

Layer	Hidden Units	Kernel Size	Stride
1	512	3	1
2	256	4	2
3	128	4	2
4	128	4	2
5	128	4	2
6	64	4	2
7	64	4	2
8	64	4	2
9	32	4	2
10	32	4	2
11	32	4	2
12	16	4	2
13	16	4	2
14	16	4	2

Notes. All layers use ConvTranspose1D with LeakyReLU(0.1) as the activation function. Hidden units for layer i are its chanels_in and chanels_out for layer i + 1. For example, layer 6 has chanels_in = 128 and chanels_out = 64.

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.

Layer	Hidden Units	Kernel Size	Stride
1	512	3	1
2	256	4	2
3	128	4	2
4	128	4	2
5	128	4	2
6	64	4	2
7	64	4	2
8	64	4	2
9	32	4	2
10	32	4	2
11	32	4	2
12	16	4	2
13	16	4	2
14	16	4	2

Layer	Hidden Units	Kernel Size	Stride
1	512	3	1
2	256	4	2
3	128	4	2
4	128	4	2
5	128	4	2
6	64	4	2
7	64	4	2
8	64	4	2
9	32	4	2
10	32	4	2
11	32	4	2
12	16	4	2
13	16	4	2
14	16	4	2

Layer	Hidden Units	Kernel Size	Stride
1	512	3	1
2	256	4	2
3	128	4	2
4	128	4	2
5	128	4	2
6	64	4	2
7	64	4	2
8	64	4	2
9	32	4	2
10	32	4	2
11	32	4	2
12	16	4	2
13	16	4	2
14	16	4	2