Table 6 Hyperparameters setting for optimal CNN training.

From: Light convolutional neural network by neural architecture search and model pruning for bearing fault diagnosis and remaining useful life prediction

Name

Value

Batch size

192

Training epochs

10

\(\omega\) Learning rate

0.025

\(\omega\) Learning rate decay

0.001

\(\omega\) Momentum

0.9

Cells count

6

Random seed

2

\(\alpha\) Grad clip

5

\(\alpha\) Learning rate

3e–4

Back to article page