Skip to main content

Official Journal of the Asia Oceania Geosciences Society (AOGS)

Table 1 Number of neurons (N) derived from NC

From: Streamlining hyperparameter optimization for radiation emulator training with automated Sherpa

 

Number of hidden layers (H)

 

1

2

3

4

5

Speedup (fold)

5

1098

392

295

248

218

(264,660)

(248,570)

(245,777)

(245,066)

(243,548)

10

549

253

196

166

147

(132,351)

(125,277)

(124,502)

(123,214)

(122,493)

20

274

158

126

109

98

(66,076)

(63,242)

(62,412)

(62,281)

(62,468)

50

110

80

68

60

54

(26,552)

(25,802)

(25,814)

(25,482)

(24,936)

100

55

45

40

36

34

(13,297)

(12,957)

(12,962)

(12,714)

(12,996)

200

27

24

22

21

20

(6549)

(6426)

(6356)

(6489)

(6542)

500

11

10

10

10

9

(2693)

(2562)

(2672)

(2782)

(2571)

1000

5

5

5

5

5

(1247)

(1277)

(1307)

(1337)

(1367)

  1. Parentheses show the NC of the given mean input (I = 175), output (O = 42), and the number of hidden layers (H)