Table 1 Definitions of symbols.
From: The impact of CNN MHAM-enhanced WRF and BPNN models for user behavior prediction
Symbol | Definition |
|---|---|
\(\:p\) | Weighted sum of the purity of non-leaf nodes in the decision tree |
\(\:imp\) | Feature importance calculated using the gradient boosting tree method |
\(\:dist\) | Distribution of sample categories within each leaf node |
\(\:{w}_{i}\) | Weight of sample ii |
\(\:{d}_{i}\) | True label of sample ii |
\(\:{y}_{i}\) | Predicted probability output by the neural network for sample ii |
\(\:f\) | Activation function of the hidden layer |
\(\:g\) | Activation function of the output layer |
\(\:\eta\:\) | Learning rate |
\(\:{\delta\:}_{k}\) | Error term at the output layer |
\(\:{\delta\:}_{j}\) | Error term at the hidden layer |
\(\:{w}_{ij}\) | Weight between input layer and hidden layer |
\(\:{v}_{jk}\) | Weight between hidden layer and output layer |
\(\:{b}_{j}\) | Bias of the hidden layer |
\(\:{c}_{k}\) | Bias of the output layer |
\(\:L\) | Weighted cross-entropy loss function |
\(\:{h}_{j}\) | Output of the hidden layer |
\(\:\sum\:\) | Summation operator |
\(\:\text{s}\text{u}\text{m}({w}^{\text{*}}\cdot\:{p}_{.}node)\) | Weighted sum of purity across all non-leaf nodes in the decision tree |
