Table 2 Summary of notations.
From: Optimizing brain stroke detection with a weighted voting ensemble machine learning model
Notation | Description |
|---|---|
\(\widehat{{y}_{i}}\) | Final predicted output for the ith input sample |
\({x}_{i}\) | The ith input data sample |
\({T}_{n}{(x}_{i})\) | Predicted class label by the nth model or client for input \({x}_{i}\) |
| Summation overall N models or clients |
\(\frac{1}{N}\) | Averaging factor to compute the mean prediction from all contributors |
\(N\) | Total number of models or clients |
\(mode()\) | Statistical mode function that returns the most frequent class label |
\(L\left(\theta \right)\) | Total loss function with parameters \(\left(\theta \right)\) |
\(n\) | Total number of data samples |
\({\text{l}(\text{u}}_{i},{\widehat{u}}_{i})\) | Loss between ground truth \({\text{u}}_{i}\) and predicted output \({\widehat{u}}_{i}\) |
| Summation overall n training samples |
| Summation over all K model components |
\(\Omega {(f}_{k})\) | Regularization term for the kth model component |
\({(f}_{k})\) | Model parameters of the kth component |
\(\left(\theta \right)\) | Overall set of model parameters |
\({\gamma }^{T}\) | Bias or constant term related to iteration T |
\(\lambda\) | Regularization coefficient |
\(T\) | Total number of training iterations or time steps |
\({w}_{j}^{2}\) | Model weight parameter at step j |
\(\sum_{j=1}^{T}{w}_{j}^{2}\) | Sum of squared weights |
\(\frac{1}{2}\lambda \sum_{j=1}^{T}{w}_{j}^{2}\) | L2 regularization term |
\({f}_{k} {(x}_{i})\) | Output of the kth model when applied to input \({x}_{i}\) |
\(K\) | Total number of models contributing to the aggregation |
| Summation over all K models |
\({\widehat{Y}}_{k}\) | An estimated value at index k |
| Summation from j = 0 to j = n, so summing up n + 1 terms |
\({Y}_{k}^{\left(j\right)}\) | Classifier |
\({w}_{j}\) | Weight assigned to jth classifier |
