Table 2 Baseline results for cancer detection, using a Fully Connected Neural Network (the classification architecture, without the AE as top layers)
From: Using autoencoders as a weight initialization method on deep neural networks for disease detection
Accuracy (%) | MCC | Precision (%) | Recall (%) | F1 score | |
|---|---|---|---|---|---|
Thyroid | 80.00 ±10.92 | 0.23 ±0.13 | 55.42 ±32.18 | 26.07 ±27.31 | 26.04 ±16.31 |
Skin | 84.67 ±4.95 | 0.46 ±0.12 | 62.08 ±20.86 | 54.15 ±20.82 | 51.15 ±14.76 |
Stomach | 81.33 ±11.47 | 0.27 ±0.18 | 56.66 ±32.34 | 33.85 ±27.06 | 30.94 ±17.66 |
Breast | 85.60 ±2.07 | 0.34 ±0.16 | 80.95 ±21.33 | 25.60 ±17.51 | 33.99 ±19.01 |
Lung | 77.13 ±12.59 | 0.21 ±0.18 | 39.90 ±37.04 | 33.73 ±35.76 | 25.25 ±22.47 |