Classification of epileptic seizure using rotation forest ensemble method with 1D-LBP feature extraction

In this article, an approach for extracting and choosing aspects of unprocessed EEG signals using
the one-dimensional local binary pattern (1D-LBP) is presented. In view of making the right
decision on different issues, notably in the medical applications like epilepsy diagnosis, a
functional approach to extracting the best aspects of EEG signals has been developed. There are
two major steps in the suggested method: First, features are extracted and selected using
enhanced 1D-LBP model that combines 1D-LBP adjacent models with PCA (1D-LBPc2p). The
second step involves classification using random rotating forest ensemble classification
algorithm. A comparison of the proposed method with 13 different reported methodologies,
comprising uniform and non-uniform 1D-LBP, is carried out. In comparison to other models
and classifiers, the results show that the technique described in proposed method has high
detection rate as well as competence by giving high accuracy. The method provided in this paper
can be used to extract features and pick EEG signals and data sets from a variety of sources.