Mixture of autoregressive modeling orders and its implication on single trial EEG classification

Abstract

Autoregressive (AR) models are of commonly utilized feature types in Electroencephalogram (EEG) studies due to offering better resolution, smoother spectra and being applicable to short segments of data. Identifying correct AR’s modeling order is an open challenge. Lower model orders poorly represent the signal while higher orders increase noise. Conventional methods for estimating modeling order includes Akaike Information Criterion (AIC), Bayesian Information Criterion (BIC) and Final Prediction Error (FPE). This article assesses the hypothesis that appropriate mixture of multiple AR orders is likely to better represent the true signal compared to any single order. Better spectral representation of underlying EEG patterns can increase utility of AR features in Brain Computer Interface (BCI) systems by increasing timely & correctly responsiveness of such systems to operator’s thoughts. Two mechanisms of Evolutionary-based fusion and Ensemble-based mixture are utilized for identifying such appropriate mixture of modeling orders. The classification performance of the resultant AR-mixtures are assessed against several conventional methods utilized by the community including 1) A well-known set of commonly used orders suggested by the literature, 2) conventional order estimation approaches (e.g., AIC, BIC and FPE), 3) blind mixture of AR features originated from a range of well-known orders. Five datasets from BCI competition III that contain 2, 3 and 4 motor imagery tasks are considered for the assessment. The results indicate superiority of Ensemble-based modeling order mixture and evolutionary-based order fusion methods within all datasets.

Non-linear classifiers applied to EEG analysis for epilepsy seizure detection

“…entropies Acharya et al (2015), energy distribution Omerhodzic et al (2013); Orhan et al (2011); Patnaik & Manyam (2008); quantitative statistical variables such as the mean, standard derivation, variance, inter-quartile range and other measurements Pippa et al (2015); autoregressive models (AR) Atyabi et al (2016); Chen (2014); or independent component analysis Siuly & Li (2015), just to name some of the most promising approaches. The type and number of such features has a direct impact in the behaviour of the system.…”

Time-varying effective EEG source connectivity: the optimization of model parameters

“…The model order selection in tv-MVAR models is not trivial [7] and has been usually performed by means of information-based criteria that neglect the non-stationary nature of the signals (e.g., Akaike information criterion, AIC [8], and Bayesian information criterion, BIC [9]) [10] [11] or through modified time-varying versions of the same criteria (e.g., the modified AIC, MAIC [12]). Other approaches have relied on choosing the model order a priori [13] or based on a comparison between parametric and non-parametric spectral density estimates [14] [15], thus, considering only the univariate part of the tv-MVAR system.…”

Intrusion detection system using Online Sequence Extreme Learning Machine (OS-ELM) in advanced metering infrastructure of smart grid

“…To deal with Distributed Denial of Service (DDoS) attack on the AMI network. [ 13 ] introduces honey into the AMI network as a decoy system to detect and gather attack information. Ntalampiras [ 14 ] proposes a novel methodology for automatic identification of integrity attacks and applies the approach to data coming from the IEEE-9 bus model, In addition, he proposed an anomaly-based methodology for reliable detection of integrity attacks in cyber-physical critical infrastructures in [ 15 ].…”

“…The research of network intrusion detection technology is developing rapidly, existing work such as [ 6 ], [ 12 ], [ 13 ] is closely related to our work. However, Zhang et al [ 12 ] focus on the intrusion detection in complete smart grid rather than aim at the security in AMI.…”

“…However, Zhang et al [ 12 ] focus on the intrusion detection in complete smart grid rather than aim at the security in AMI. Wang et al [ 13 ] pay more attention to DDoS which is just one type of attack types, whereas we concentrate on performance and effectiveness with various of attack types in whole IDS. Zhang et al did not give a specific comparison of used classifiers, which is outmoded.…”