EEG Signal Classification Based On Iterative Random Forest Algorithm

With the rapid development of data science and biomedicine,the research based on eeg and other Electrophysiological signals has attracted great attention in the field of machine learning.Electroencephalogram(EEG)is one of the most important electrophysiological signals in human body.EEG can accurately reflect human brain consciousness and brain health.It is an important and challenging research to discover the important features of EEG signals and to construct an effective method of EEG signal classification.In this paper,EEG signal classification method based on local mean decomposition and iterative random forest algorithm is proposed.The local mean decomposition method(LMD)has a certain degree of adaptability,which is less than some traditional signal processing methods in terms of mode aliasing.Therefore,this paper adopts this method to conduct signal processing for different EEG signals.In this paper,a method of EEG signal analysis combining local mean decomposition and iterative random forest is proposed.300 single channel signals were selected from the EEG data set of Bonn University,which were collected from the scalp surface layer of normal people when they opened their eyes,and from the epileptogenic area of epileptic patients during the seizure and the epileptogenic area between the seizures.Firstly,we use matlab to realize the local mean decomposition and process the EEG signals.After that,SAS is used to extract the feature of product function and trend term.Three machine learning methods,i.e.iterative random forest,random forest and support vector machine,are realized through R respectively.The optimal parameters are adjusted.The cross validation of 10 fold is used for each method.The performance of the classification model is evaluated by comparing the classification results of several methods,and the method with the highest classification accuracy is selected.The experimental results show that the algorithm proposed in this paper,which combines the adaptive decomposition and the iterative random forest,has the best effect in the EEG data classification of epileptic patients in the seizure period and the interval period,and achieves the classification accuracy of 99.14%.The accuracy of multiple classification between the EEG signal and the seizure stage and the interseizure stage of healthy volunteers reached 75.05%,The classification accuracy is higher than that of random forest and support vector machine.In conclusion,this paper takes EEG signal as the research object,discusses the core EEG signal classification algorithm in the automatic detection technology of epilepsy,and uses the method of local mean decomposition and iterative random forest to carry out accurate and stable epilepsy detection.This paper can promote the development of EEG signal automatic detection,and provide an effective and feasible method for accurate identification of epileptic EEG signal.