From: Flight delay prediction based on deep learning and Levenberg-Marquart algorithm
Notation | Description | Notation | Description |
---|---|---|---|
X | Input matrix | \({\hat{\text{X}}}\) | Reconstruct-ed \({\tilde{\text{X}}}\) |
Min(x) | Lowest value in series and its number is zero | WT | Transposition of the weight matrix W |
Max(x) | Highest number in series and has value of 1 | bh | Show the bias associated with each hidden code |
\({\text{X}}_{{\text{i}}}^{1}\) | Normalized data in first layer | L (X, \({\tilde{\text{X}}}\)) | Reconstruct-ion error rate |
h | Hidden layer | cost | Error rate |
\({\tilde{\text{X}}}\) | Corrupted input | W | Weighted matrix |
c | Corruption level | y | Output per x |
H | Activation function | \({\hat{\text{y}}}\) | Output per \({\hat{\text{X}}}\) |
\({\text{O}}_{{{\text{j}} - 1}} \to {\text{O}}_{{\text{j}}}\) | Each layer’s input is from previous layer’s output | θ | Is the parameters of the denoising autoencoder |
b | Bias vector | Â | Â |