Читать книгу Smart Healthcare System Design - Группа авторов - Страница 32

In order to find the features with the most potential, an algorithm was implemented to approximate individual feature strength with respect to every other feature [30]. The strength of a feature was determined by the accuracy with which the preictal state was classified as an average of several classifications. Similar to Cross-Validation by Elimination HANNSVM algorithm repartitions the feature set, performs a set of classifications, finds the best feature sets to drop, and then adjusts the feature space to only contain features that improve the accuracy.

1 1. Evaluate the accuracy of the classification using all N feature sets.

2 2. Dropping one feature set at a time, repartitions the feature space into N, N − 1 feature subsets and save the accuracy of each sub set at position K in vector P along with the resulting accuracy.

3 3. Denote the index of P with the maximum accuracy as B, and drop all the features listed in P from B to N from the final feature space.

The resulting feature set P has accuracy similar to the accuracy found at position B in P. Under training and overtraining must still be taken into consideration since it can have an effect on the accuracy of a prediction.