Electroencephalogram (EEG) signal analysis plays an essential role in detecting and understanding epileptic seizures. It is known that seizure processes are nonlinear and non-stationary, discriminating between rhythmic discharges and dynamic change is a challenging task in EEG-based seizure detection. In this paper, a new time-varying (TV) modeling framework, based on an autoregressive (AR) model structure, is proposed to characterize and analyze EEG signals. The TV parameters of the AR model are approximated through a multi-wavelet basis function expansion (MWBF) approach. An effective ultra-regularized orthogonal forward regression (UROFR) algorithm is employed to significantly reduce and refine the resulting expanded model. Given a time-varying process, the proposed TVAR–MWBF–UROFR method can generate a parsimonious TVAR model, based on which a high-resolution power spectrum density (PSD) estimation can be obtained. Informative features are then defined and extracted from the PSD estimation. The TVAR–MWBF–UROFR method is applied to a number of real EEG datasets; features obtained from these datasets are then used for seizure detection and classification. To make the results more accurate and reliable, a PCA algorithm is adopted to select the optimal feature subset, and a Bayesian optimization technique based on the Gaussian process is performed to determine the coefficients associated with each of the classifiers. The performance of the proposed method is tested on two benchmark datasets, and the experimental results indicate that TVAR–MWBF–UROFR outperforms the compared state-of-the-art classifiers in terms of accuracy, specificity, sensitivity and robustness.

脑电图（EEG）信号分析在检测和了解癫痫发作中起着至关重要的作用。众所周知，癫痫发作过程是非线性且不稳定的，在基于EEG的癫痫发作检测中区分节奏性放电和动态变化是一项艰巨的任务。本文提出了一种基于自回归（AR）模型结构的时变（TV）建模框架，用于表征和分析脑电信号。AR模型的电视参数是通过多小波基函数展开（MWBF）方法估算的。有效的超正则化正交正向回归（UROFR）算法用于显着减少和完善所得扩展模型。给定随时间变化的过程，建议的TVAR–MWBF–UROFR方法可以生成简约的TVAR模型，基于此，可以获得高分辨率功率谱密度（PSD）估计。然后定义信息特征，并从PSD估计中提取信息特征。TVAR–MWBF–UROFR方法适用于许多真实的EEG数据集；然后将从这些数据集中获得的特征用于癫痫发作检测和分类。为了使结果更准确和可靠，采用PCA算法选择最优特征子集，并基于高斯过程进行贝叶斯优化技术，以确定与每个分类器相关的系数。在两个基准数据集上测试了该方法的性能，实验结果表明，TVAR–MWBF–UROFR在准确性，特异性，灵敏度和鲁棒性方面均优于同类的最新分类器。然后定义信息特征，并从PSD估计中提取信息特征。TVAR–MWBF–UROFR方法适用于许多真实的EEG数据集；然后将从这些数据集中获得的特征用于癫痫发作检测和分类。为了使结果更准确和可靠，采用PCA算法选择最优特征子集，并基于高斯过程进行贝叶斯优化技术，以确定与每个分类器相关的系数。在两个基准数据集上测试了该方法的性能，实验结果表明，TVAR–MWBF–UROFR在准确性，特异性，灵敏度和鲁棒性方面均优于同类的最新分类器。然后定义信息特征，并从PSD估计中提取信息特征。TVAR–MWBF–UROFR方法适用于许多真实的EEG数据集；然后将从这些数据集中获得的特征用于癫痫发作检测和分类。为了使结果更准确和可靠，采用PCA算法选择最优特征子集，并基于高斯过程进行贝叶斯优化技术，以确定与每个分类器相关的系数。在两个基准数据集上测试了该方法的性能，实验结果表明，TVAR–MWBF–UROFR在准确性，特异性，灵敏度和鲁棒性方面均优于同类的最新分类器。然后将从这些数据集中获得的特征用于癫痫发作检测和分类。为了使结果更准确和可靠，采用PCA算法选择最优特征子集，并基于高斯过程进行贝叶斯优化技术，以确定与每个分类器相关的系数。在两个基准数据集上测试了该方法的性能，实验结果表明，TVAR–MWBF–UROFR在准确性，特异性，灵敏度和鲁棒性方面均优于同类的最新分类器。然后将从这些数据集中获得的特征用于癫痫发作检测和分类。为了使结果更准确和可靠，采用PCA算法选择最优特征子集，并基于高斯过程进行贝叶斯优化技术，以确定与每个分类器相关的系数。在两个基准数据集上测试了该方法的性能，实验结果表明，TVAR–MWBF–UROFR在准确性，特异性，灵敏度和鲁棒性方面均优于同类的最新分类器。执行基于高斯过程的贝叶斯优化技术，以确定与每个分类器相关的系数。在两个基准数据集上测试了该方法的性能，实验结果表明，TVAR–MWBF–UROFR在准确性，特异性，灵敏度和鲁棒性方面均优于同类的最新分类器。执行基于高斯过程的贝叶斯优化技术，以确定与每个分类器相关的系数。在两个基准数据集上测试了该方法的性能，实验结果表明，TVAR–MWBF–UROFR在准确性，特异性，灵敏度和鲁棒性方面均优于同类的最新分类器。