Wavelet-package transformation as a preprocessor of EEG waveforms for classification

Proceedings of the 19th Annual International Conference of the IEEE Engineering in Medicine and Biology Society. 'Magnificent Milestones and Emerging Opportunities in Medical Engineering' (Cat. No.97CH36136) Pub Date : 1997-10-30 DOI:10.1109/IEMBS.1997.756615

Gdmundur Saevarsson, J. R. Sveinsson, J. Benediktsson

{"title":"Wavelet-package transformation as a preprocessor of EEG waveforms for classification","authors":"Gdmundur Saevarsson, J. R. Sveinsson, J. Benediktsson","doi":"10.1109/IEMBS.1997.756615","DOIUrl":null,"url":null,"abstract":"The results of this paper show that preprocessing of an EEG signal with wavelet packet transformation, both enhances the feature detection capability of a classifier and reduces the input vectors dimensions considerably. The best basis method gave perfect classification with the hold-out method and would be considered to be the best method used in the experiment. It shows that the selection of the packets for the feature vector can be selected with best basis criterions like the minimum entropy criteria. There are few things though that could explain this results. First the results are one shot results, a process like Monte Carlo was not used mainly because of low availability of training samples. The results are not either an average of random selections for the training and test samples, so the way the samples were split up could make a difference. Wavelet packet transformation has shown itself to be a powerful tool in preprocessing of feature vectors for classification. The classifier does not have to be statistical, it could also be a neural network or any other pattern recognition system.","PeriodicalId":342750,"journal":{"name":"Proceedings of the 19th Annual International Conference of the IEEE Engineering in Medicine and Biology Society. 'Magnificent Milestones and Emerging Opportunities in Medical Engineering' (Cat. No.97CH36136)","volume":"385 1","pages":"0"},"PeriodicalIF":0.0000,"publicationDate":"1997-10-30","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":"20","resultStr":null,"platform":"Semanticscholar","paperid":null,"PeriodicalName":"Proceedings of the 19th Annual International Conference of the IEEE Engineering in Medicine and Biology Society. 'Magnificent Milestones and Emerging Opportunities in Medical Engineering' (Cat. No.97CH36136)","FirstCategoryId":"1085","ListUrlMain":"https://doi.org/10.1109/IEMBS.1997.756615","RegionNum":0,"RegionCategory":null,"ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":null,"EPubDate":"","PubModel":"","JCR":"","JCRName":"","Score":null,"Total":0}

引用次数: 20

Abstract

The results of this paper show that preprocessing of an EEG signal with wavelet packet transformation, both enhances the feature detection capability of a classifier and reduces the input vectors dimensions considerably. The best basis method gave perfect classification with the hold-out method and would be considered to be the best method used in the experiment. It shows that the selection of the packets for the feature vector can be selected with best basis criterions like the minimum entropy criteria. There are few things though that could explain this results. First the results are one shot results, a process like Monte Carlo was not used mainly because of low availability of training samples. The results are not either an average of random selections for the training and test samples, so the way the samples were split up could make a difference. Wavelet packet transformation has shown itself to be a powerful tool in preprocessing of feature vectors for classification. The classifier does not have to be statistical, it could also be a neural network or any other pattern recognition system.

查看原文本刊更多论文

基于小波包变换的脑电信号分类预处理方法

研究结果表明，采用小波包变换对脑电信号进行预处理，既提高了分类器的特征检测能力，又大大降低了输入向量的维数。最佳基法与hold-out法具有较好的分类效果，是实验中使用的最佳基法。结果表明，可以用最小熵准则等最佳基准则来选择特征向量的包。然而，几乎没有什么可以解释这一结果。首先，结果是一次性的结果，没有使用像蒙特卡罗这样的过程，主要是因为训练样本的可用性低。结果既不是训练样本和测试样本随机选择的平均值，所以样本分割的方式可能会有所不同。小波包变换在特征向量预处理分类中已成为一种强有力的工具。分类器不一定是统计的，它也可以是神经网络或任何其他模式识别系统。

本文章由计算机程序翻译，如有差异，请以英文原文为准。

求助全文

约1分钟内获得全文求助全文

来源期刊

Proceedings of the 19th Annual International Conference of the IEEE Engineering in Medicine and Biology Society. 'Magnificent Milestones and Emerging Opportunities in Medical Engineering' (Cat. No.97CH36136)

自引率

0.00%

发文量