Table 4 The trained forward pass performance. The default float type is float32 on the GPU. The measures are defined as follows: \(\mbox{MSE} = \frac{1}{n} \sum (y_{i}-\hat{y}_{i})^{2}\), \(\mbox{MAE} = \frac{1}{n} \sum |y_{i}-\hat{y}_{i}|\), \(\mbox{MAPE} = \frac{1}{n} \sum \frac{|y_{i}-\hat{y}_{i}|}{y_{i}}\), where y represents the true value, and ŷ represents the predicted value with n being the number of samples
From: A neural network-based framework for financial model calibration
Heston–IV–ANN | MSE | MAE | MAPE | \(R^{2}\) |
|---|---|---|---|---|
Training | 8.07 × 10−8 | 2.15 × 10−4 | 5.83 × 10−4 | 0.9999936 |
Testing | 1.23 × 10−7 | 2.40 × 10−4 | 7.20 × 10−4 | 0.9999903 |