Abstract: In order to achieve effective identification of EEG( Electroencephalogram) signals in patients with
epilepsy，improve the quality of life for patients，a method of EEG signal classification based on the combination
of local mean decomposition and iterative random forest is proposed for the non-stationary and nonlinear
characteristics of EEG signals． Firstly，the EEG signal is decomposed into several product function components
and a residual component by using local mean decomposition． Then all components are extracted and classified
using support vector machine，random forest and iterative random forest methods． The experimental results show
that the classification accuracy of iterative random forest is higher than that of support vector machine and random
forest method． This method provides a feasible and effective way to accurately identify epileptic EEG signals，and
has good application value．

Key words: electroencephalogram ( EEG) signal, feature extraction, local mean decomposition, iterative
random forest