Table 7 Experimented autoencoder structures with different combinations of parameters
From: Machine learning to analyse omic-data for COVID-19 diagnosis and prognosis
Parameter | Model 1 | Model 2 | Model 3 | Model 4 |
|---|---|---|---|---|
Number of layers | 5 | 5 | 3 | 7 |
Connect layer size | 2000 | 1000 | 1000 | 3000 |
Bottleneck size | 500 | 500 | 300 | 500 |
Learning rate | 0.001 | 0.001 | 0.0001 | 0.0001 |
Epoch | 10 | 10 | 20 | 20 |
Dropout | 0.2 | 0.5 | 0.2 | 0.5 |
c-index | 0.76 | 0.68 | 0.62 | 0.59 |