M. Kamrunnahar

Feature selection on movement imagery discrimination and attention detection

Noninvasive brain–computer interfaces (BCI) translate subject’s electroencephalogram (EEG) features into device commands. Large feature sets should be down-selected for efficient feature translation. This work proposes two different feature down-selection algorithms for BCI: (a) a sequential forward selection; and (b) an across-group variance. Power rar ratios (PRs) were extracted from the EEG data for movement imagery discrimination. Event-related potentials (ERPs) were employed in the discrimination of cue-evoked responses. While center-out arrows, commonly used in calibration sessions, cued the subjects in the first experiment (for both PR and ERP analyses), less stimulating arrows that were centered in the visual field were employed in the second experiment (for ERP analysis). The proposed algorithms outperformed other three popular feature selection algorithms in movement imagery discrimination. In the first experiment, both algorithms achieved classification errors as low as 12.5% reducing the feature set dimensionality by more than 90%. The classification accuracy of ERPs dropped in the second experiment since centered cues reduced the amplitude of cue-evoked ERPs. The two proposed algorithms effectively reduced feature dimensionality while increasing movement imagery discrimination and detected cue-evoked ERPs that reflect subject attention.

Estimation of Markov parameters and time-delay/interactor matrix

Abstract The ARMarkov-least squares method is extended to multivariable systems. This method explicitly determines the Markov parameters (impulse response coefficients) of a process using process input–output data and a standard least-squares algorithm. The parameter estimates are consistent and have tighter confidence bounds than those produced by other linear regression methods. The Interactor matrix, which defines the time delays in multivariable systems, can be directly estimated from the Markov parameters. From simulation results it is observed that the Markov parameters estimated by the ARMarkov-LS method are the closest to the actual Markov parameters irrespective of the system order and lead to a better estimate of the interactor matrix than other linear regression methods such as Correlation Analysis, ARX, FIR, etc. The identified Markov parameters and/or the time-delay/interactor matrix can be used directly in the design of model predictive controllers and control loop performance assessment.

Parameter Sensitivity Analysis of Pit Initiation at Single Sulfide Inclusions in Stainless Steel

Sensitivity analysis methods were used in conjunction with a mathematical model for corrosion pit initiation in the vicinity of MnS inclusions in stainless steel to investigate the relationship between physicochemical parameters and the potential and concentration distributions. The finite difference method with central differences was used to calculate sensitivities. The mathematical model of pit initiation included 20 species plus the potential and 13 physicochemical parameters including rate constants for chemical and electrochemical surface reactions and equilibrium constants for homogeneous reactions. It was found that the potential and concentration profiles are most sensitive to the Tafel slope of the rate of electrochemical dissolution of sulfur-containing inclusions and least sensitive to changes in the equilibrium coefficients of the homogeneous reactions. The rate constant for the electrochemical reaction for dissolution of sulfide inclusions was also found to be significant. The procedure provides a first step toward selecting the most important parameters, designing critical experiments, and selecting the hypothesis that best fits experimental data. Pitting corrosion of stainless steel ~SS! is a localized phenomenon that may initiate at various types of surface sites including sulfide inclusions. Interest in pitting corrosion is high because it is often a first step leading to crevice corrosion, corrosion fatigue, stress-corrosion cracking, and failure of coatings. The various mechanisms by which initiation occurs, a subject of longstanding interest, have increasingly been investigated with the aid of mathematical models. While there are various modeling approaches, we focus here on an approach where the underlying physical phenomena associated with the mechanisms are expressed by continuum equations for reaction, transport, and equilibration among species. Although numerical simulation of complex corrosion systems can provide useful insight, there is also the need for additional numerical methods. In this work we focus on the assessment of uncertainty. For example, the validation of models by comparison with experimental data requires specification of the hypothesis of corrosion mechanism ~of which the literature provides multiple reasonable choices! as well as values for the system parameters ~some of which may be difficult or impossible to measure directly!. Various kinds of uncertainty therefore arise. The motivation for the present work is to apply numerical analysis tools to identify the most sensitive parameters associated with one particular hypothesis of mechanism. Such tools may find use in addressing questions such as: What properties of a system are responsible for its observed behavior? What is the most promising experiment to refute or confirm a model? Which of several hypotheses best agrees with experimental data from heterogeneous sources? The role of sulfide inclusions has been widely investigated. Sulfide inclusions play an important role in the initiation of pitting corrosion. Various researchers have studied initiation of pitting corrosion with a range of experimental techniques to clarify various events during early stages of sulfide inclusion dissolution, ~e.g., Ref. 1-3! and pit growth. Sulfur-containing species have been detected during dissolution of sulfide inclusions, 4-8 and pH measurements have been taken at various locations during sulfide dissolution. 9 The influence of applied mechanical stress on pit initiation has been investigated. 10 More recently, an electrochemical microcell was used to obtain electrochemical data and chemical composition at the vicinity of the sulfide inclusions. 11,12 In the present work we consider in detail a mathematical model of one particular mechanism developed to simulate pit initiation at a single MnS inclusion in SS within an electrochemical microcell. 13 The model examined the hypotheses that pit initiation occurs by depassivation of SS as a result of accumulation of thiosulfate ions above a critical concentration in the presence of chloride, and that the rate of inclusion dissolution was catalyzed by chloride. The model was used to predict the variation of potential in time and distance during the pit initiation and also to predict the dependence of pitting potential on the chloride concentration for a single inclusion. The emphasis in the present work is to apply numerical procedures for assessment of parameter sensitivity and to demonstrate their use with one hypothesis of mechanism. In this work, the system of coupled nonlinear equations reported

Comparison of EEG Pattern Classification Methods for Brain-Computer Interfaces

The aim of this study is to compare 2 EEG pattern classification methods towards the development of BCI. The methods are: (1) discriminant stepwise, and (2) principal component analysis (PCA) - linear discriminant analysis (LDA) joint method. Both methods use Fishers LDA approach, but differ in the data dimensionality reduction procedure. Data were recorded from 3 male subjects 20-30 years old. Three runs per subject took place. The classification methods were tested in 240 trials per subject after merging all runs for the same subject. The mental tasks performed were feet, tongue, left hand and right hand movement imagery. In order to avoid previous assumptions on preferable channel locations and frequency ranges, 105 (21 electrodestimes5 frequency ranges) electroencephalogram (EEG) features were extracted from the data. The best performance for each classification method was taken into account. The discriminant stepwise method showed better performance than the PCA based method. The classification error by the stepwise method varied between 31.73% and 38.5% for all subjects whereas the error range using the PCA based method was 39.42% to 54%.

Optimization of electrode channels in brain computer interfaces

What is the optimal number of electrodes one can use in discrimination of tasks for a Brain Computer Interface (BCI)? To address this question, the number and location of scalp electrodes in the acquisition of human electroencephalography (EEG) and discrimination of motor imagery tasks were optimized by using a systematic optimization approach. The systematic analysis results in the most reliable procedure in electrode optimization as well as a validating means for the other feature selection techniques. We acquired human scalp EEG in response to cue-based motor imagery tasks. We employed a systematic analysis by using all possible combinations of the channels and calculating task discrimination errors for each of these combinations by using linear discriminant analysis (LDA) for feature classification. Channel combination that resulted in the smallest discrimination error was selected as the optimum number of channels to be used in BCI applications. Results from the systematic analysis were compared with another feature selection algorithm: forward stepwise feature selection combined with LDA feature classification. Our results demonstrate the usefulness of the fully optimized technique for a reliable selection of scalp electrodes in BCI applications.

Model predictive control using an extended ARMarkov model

Abstract The original ARMarkov identification method explicitly determines the first μ Markov parameters from plant input–output data and approximates the slower dynamics of the process by an ARX model structure. In this paper, the method is extended to include a disturbance model and an ARIMAX structure is used to approximate the slower dynamics. This extended ARMarkov model is then used to formulate a predictive controller. As the number of Markov parameters in the model varies from one to P (prediction horizon)+1, the controller changes from generalized predictive control (GPC) to dynamic matrix control (DMC). The advantages of the proposed ARM-MPC are the consistency of the Markov parameters estimated by the ARMarkov method, independent tuning of the controller for servo and regulatory responses and the ability to combine the characteristics of GPC and DMC. The theoretical results are illustrated through simulation examples.

Toward a Model-Based Predictive Controller Design in Brain–Computer Interfaces

A first step in designing a robust and optimal model-based predictive controller (MPC) for brain–computer interface (BCI) applications is presented in this article. An MPC has the potential to achieve improved BCI performance compared to the performance achieved by current ad hoc, nonmodel-based filter applications. The parameters in designing the controller were extracted as model-based features from motor imagery task-related human scalp electroencephalography. Although the parameters can be generated from any model-linear or non-linear, we here adopted a simple autoregressive model that has well-established applications in BCI task discriminations. It was shown that the parameters generated for the controller design can as well be used for motor imagery task discriminations with performance (with 8–23% task discrimination errors) comparable to the discrimination performance of the commonly used features such as frequency specific band powers and the AR model parameters directly used. An optimal MPC has significant implications for high performance BCI applications.

Model-based responses and features in Brain Computer Interfaces

Novel model based features are introduced in the discrimination of motor imagery tasks using human scalp electroencephalography (EEG) towards the development of Brain Computer Interfaces (BCI). We have acquired human scalp EEG under open-loop and feedback conditions in response to cue-based motor imagery tasks. EEG signals, transformed into frequency specific bands such as mu, beta and movement related potentials, were used for feature extraction with the aim to discriminate tasks. Data were classified using features such as power spectrum and model-based parameters. Two different feature selection methods: stepwise and principal component analysis (PCA), were combined with linear discriminant analysis (LDA). Different training/validation criteria were applied for classification of task related features. Results show that the scalp EEG correlate of the imagery tasks of hands/toes/tongue movements under open-loop conditions and left/right hand movements under feedback conditions, can be well discriminated with classification errors below 20%. Model based techniques, which resulted in classification errors in the range of 2%–30%, have the potential to use advanced control systems theory in the development of BCI to achieve improved performance compared to the performance achieved by currently applied proportional control or filter algorithms.

A simple generative model applied to motor-imagery brain-computer interfacing

In this study, a generative model is developed in order to translate neural activity into predictable device commands for brain-computer interface (BCI) applications. Generative approaches to BCI translation differ from widely-used discriminative approaches because they develop a model of brain activity dependent on the mental state of the user. Preliminary results indicate that two of three subjects were able to control the system at a level (>;70% accurate) that makes it a viable option for practical use. The accuracy rate of the generative model is compared to the accuracy rate calculated offline using a linear discriminant approach. The advantages of such a system are discussed, and the ongoing opportunities for paradigm improvement are outlined.

Visual evoked potentials for attentional gating in a brain-computer interface

For synchronous brain-computer interface (BCI) paradigms tasks that utilize visual cues to direct the user, the neural signals extracted by the computer are representative of voluntary modulation as well as evoked responses. For these paradigms, the evoked potential is often overlooked as a source of artifact. In this paper, we put forth the hypothesis that cue priming, as a mechanism for attentional gating, is predictive of motor imagery performance, and thus a viable option for self-paced (asynchronous) BCI applications. We approximate attention by the amplitude features of visually evoked potentials (VEP)s found using two methods: trial matching to an average VEP template, and component matching to a VEP template defined using independent component analysis (ICA). Templates were used to rank trials that display high vs. low levels of fixation. Our results show that subject fixation, measured by VEP response, fails as a predictor of successful motor-imagery task completion. The implications for the BCI community and the possibilities for alternative cueing methods are given in the conclusions.