No full-text available
To read the full-text of this research,
you can request a copy directly from the authors.
Asymmetric hemisphere modeling in an offline brain-computer interface
Abstract
Classification of the electroencephalogram (EEG) during motor
imagery of the left or right hand can be performed using a classifier
comprising two hidden Markov models (HMMs) describing the
spatio-temporal patterns related to the imagination. Due to the known
asymmetries during motor imagery of rightand left-hand movement, an
HMM-based classifier allowing asymmetrical structures is introduced. The
comparison between such a system and a symmetrical one is based on the
error rate of classification. The results for EEG data collected during
20 sessions from five subjects demonstrate a significant improvement of
9% for the classification accuracy for the asymmetric classifiers. The
selection of the DAM for classification is done using a variant of
genetic algorithms (GAs); namely, the adaptive reservoir genetic
algorithm (ARGA)
... Reference [40] studied motor imagery performance after asymmetrical transcranial direct current stimulation. Reference [41] introduced an HMM-based classifier allowing asymmetrical structures to help design the EEG-based BCI training paradigm. Due to the known asymmetries during motor imagery of right-and left-hand movement, the classifier of imagery tasks demonstrates an improvement of 9% for the classification accuracy. ...
It is essential for neuroscience and clinic to estimate the influence of neuro-intervention after brain damage. Most related studies have used Mirrored Contralesional-Ipsilesional hemispheres (MCI) methods flipping the axial neuroimaging on the x-axis in prognosis prediction. But left-right hemispheric asymmetry in the brain has become a consensus. MCI confounds the intrinsic brain asymmetry with the asymmetry caused by unilateral damage, leading to questions about the reliability of the results and difficulties in physiological explanations. We proposed the Separated Left-Right hemiplegia (SLR) method to model left and right hemiplegia separately. Two pipelines have been designed in contradistinction to demonstrate the validity of the SLR method, including MCI and removing intrinsic asymmetry (RIA) pipelines. A patient dataset with 18 left-hemiplegic and 22 right-hemiplegic stroke patients and a healthy dataset with 40 subjects, age- and sex-matched with the patients, were selected in the experiment. Blood-Oxygen Level-Dependent MRI and Diffusion Tensor Imaging were used to build brain networks whose nodes were defined by the Automated Anatomical Labeling atlas. We applied the same statistical and machine learning framework for all pipelines, logistic regression, artificial neural network, and support vector machine for classifying the patients who are significant or non-significant responders to brain-computer interfaces assisted training and optimal subset regression, support vector regression for predicting post-intervention outcomes. The SLR pipeline showed 5-15% improvement in accuracy and at least 0.1 upgrades in R
2
, revealing common and unique recovery mechanisms after left and right strokes and helping clinicians make rehabilitation plans.
... Sonuçlar göstermektedir ki EEG sınıflandırma problemleri için dinamik sınıflandırıcılar duragan bir sınıflandırıcı olan Mahalanobis mesafe sınıflandırıcısına göre daha başarılıdırlar. Aynı şekilde ARözniteliklerinin HMMler ile dahaöncë onerilen Hjorth [9]özniteliklerinden daha iyi çalıştıkları görülmektedir. ...
We consider the problem of motor imagery EEG data classification within the context of brain-computer interfaces. We propose an approach based on Hidden Markov models (HMMs). Our approach is different from existing HMM-based techniques in that it uses features based on auto-regressive parameters together with dimensionality reduction based on principal component analysis (PCA). We demonstrate the effectiveness of our approach through experimental results for two and four-class problems based on a public dataset, as well as data collected in our laboratory.
... One approach to provide such an interface is mental task classification; i.e. each segment of EEG signal should be assigned to its appropriate class among the predefined classes of mental tasks. Different types of classifiers have been used for this purpose; such as, neural classifiers [2,[4][5][6][7][8][9] and HMM-based classifiers [10][11][12][13]. ...
... Sonuçlar göstermektedir ki EEG sınıflandırma problemleri için dinamik sınıflandırıcılar duragan bir sınıflandırıcı olan Mahalanobis mesafe sınıflandırıcısına göre daha başarılıdırlar. Aynı şekilde ARözniteliklerinin HMMler ile dahaöncë onerilen Hjorth [9]özniteliklerinden daha iyi çalıştıkları görülmektedir. ...
We consider the problem of motor imagery EEG data classification within the context of brain-computer interfaces. We propose an approach based on Hidden Markov models (HMMs). Our approach is different from existing HMM-based techniques in that it uses features based on autoregressive parameters together with dimensionality reduction based on principal component analysis (PCA). We demonstrate the effectiveness of our approach through experimental results for two and four-class problems based on a public dataset, as well as data collected in our laboratory.
... Some of the earlier methods of control have utilized power analysis of EEG signals in relation to event-related desynchronization (ERD), which is detectable upon the occurrence of blocking or attenuation of alpha rhythms [10]. ERD in the mu rhythm in the central-lateral sensorimotor area (which is simultaneous with a reported amplitude increase in the ipsilateral region) has even been observed for imagination of movement [11]. However, the lack of consistency in its occurrence among subjects and the variability of its location make dependent use of such known asymmetries unsatisfactory for robust design. ...
Real-time analysis of multi-channel spatio-temporal sensor data presents a considerable technical challenge for a number of applications. For example, in brain–computer interfaces, signal patterns originating on a time-dependent basis from an array of electrodes on the scalp (i.e. electroencephalography) must be analyzed in real time to recognize mental states and translate these to commands which control operations in a machine. In this paper we describe a new technique for recognition of spatio-temporal patterns based on performing online discrimination of time-resolved events through the use of correlation of phase dynamics between various channels in a multi-channel system. The algorithm extracts unique sensor signature patterns associated with each event during a training period and ranks importance of sensor pairs in order to distinguish between time-resolved stimuli to which the system may be exposed during real-time operation. We apply the algorithm to electroencephalographic signals obtained from subjects tested in the neurophysiology laboratories at the University of Toronto. The extension of this algorithm for rapid detection of patterns in other sensing applications, including chemical identification via chemical or bio-chemical sensor arrays, is also discussed.
... The algorithm builds on a former variant introduced by the authors in [7] and called Adaptive Reservoir Genetic Algorithm (ARGA). The convergence within finite time, with probability 1 of ARGA, was shown to hold in [8], while a real-world application focusing on finding the best Hidden Markov Model classifier for a Brain Computer Interface task, was discussed in [10]. The present article introduces ARGAII, a variant of ARGA, that bears the basic architecture of ARGA, improving however the control mechanism by employing a Bayesian decision process. ...
It is now common knowledge that blind search algorithms cannot perform with equal efficiency on all possible optimization prob-lems defined on a domain. This knowledge applies also to Genetic Algo-rithms when viewed as global and blind optimizers. From this point of view it is necessary to design algorithms capable of adapting their search behaviour by making use in a direct fashion of the knowledge pertaining to the search landscape. The paper introduces a novel adaptive Genetic Algorithm where the exploration/exploitation balance is directly con-trolled using a Bayesian decision process. Test cases are analyzed as to how parameters affect the search behaviour of the algorithm.
... One approach to provide such an interface is mental task classification; i.e. each segment of EEG signal should be assigned to its appropriate class among the predefined classes of mental tasks. Different types of classifiers have been used for this purpose; such as, neural classifiers [2,456789 and HMM-based classifiers10111213. ...
Mental task classification using brain signals, mostly electroencephalogram (EEG), is an approach to understand human brain functions. As EEG seems to be chaotic, it is important to verify the capability of probabilistic and statistical processing tools (such as HMM-based classifiers) in working with chaotic signals. At first, we study the performance of HMM's in classification of different classes of synthetically generated chaotic signals. Then performance of such classifiers in EEG-based mental task classification is studied. Results show good performance in both cases.
... This finding also correlates with a recent study where the asymmetry of two hemispheres is modeled with a Hidden Markov Model and a genetic algorithm. The authors reported a significant improvement when such an asymmetric classifier is used for classification [15]. Fig. 1. ...
We use local cosine packets to adaptively segment EEG corresponding to left or right hand index finger movements. The segmentation is constructed by maximizing the Euclidean or Kullback-Leibler distance criterion between the left and right finger movement. The proposed method divides the movement EEG from the C3 and C4 electrodes into nonuniform time segments over a dyadic tree. The most discriminative features are selected from the pruned tree. We observed that the selected segmentation and discriminative components are subject specific. We believe this may eliminate the inter and intrasubject variability when constructing brain computer interfaces. We also found striking asymmetry between feature characteristics and their discrimination power on each hemisphere
... With respect to EEG signal classification, the applicability of many linear and nonlinear single classifiers has already been assessed, such as Fisher discriminant analysis, neural networks, support vector machines, hidden Markov models, Bayes classifiers, and source analysis (Garrett et al., 2003;Kamousi et al., 2005;Millán et al., 2004;Mü ller et al., 2003;Obermaier et al., 2001;Zhang, 2005, 2006a,b). However, as we know it may be difficult to build a good single classifier if data are of high dimensionality and the training set is comparatively small. ...
Ensemble learning for improving weak classifiers is one important direction in the current research of machine learning, and thereinto bagging, boosting and random subspace are three powerful and popular representatives. They have so far shown efficacies in many practical classification problems. However, for electroencephalogram (EEG) signal classification with application to brain–computer interfaces (BCIs), there are almost no studies investigating their feasibilities. The present study systematically evaluates the performance of the three ensemble methods for EEG signal classification of mental imagery tasks. With the base classifiers of k-nearest-neighbor, decision tree and support vector machine, classification experiments are carried out upon real EEG recordings. Experimental results suggest the feasibilities of ensemble classification methods, and we also derive some valuable conclusions on the performance of ensemble methods for EEG signal classification.
... These same 8 subjects were also used in [12] and [13]. B. Results We show the probability of correct classification (PCC) of our AR-PCA-HMM approach as compared to other HMMbased classifiers including the Hjoth-HMM approach of [14] inFigure 2. For 7 of 8 subjects, our approach achieves the highest PCC. In Table II, we present a comparison of AR- PCA-HMM with the top techniques in BCI Competition IV on this dataset 1 , in terms of the κ coefficient. ...
We propose an approach based on Hidden Markov models (HMMs) combined with principal component analysis (PCA) for classification of four-class single trial motor imagery EEG data for brain computer interfacing (BCI) purposes. We extract autoregressive (AR) parameters from EEG data and use PCA to decrease the number of features for better training of HMMs. We present experimental results demonstrating the improvements provided by our approach over an existing HMM based EEG single trial classification approach as well as over state-of-the-art classification methods.
... The algorithm builds on a former variant introduced by the authors in [7] and called Adaptive Reservoir Genetic Algorithm (ARGA). The convergence within finite time, with probability 1 of ARGA, was shown to hold in [8], while a real-world application focusing on finding the best Hidden Markov Model classifier for a Brain Computer Interface task, was discussed in [10]. The present article introduces ARGAII, a variant of ARGA, that bears the basic architecture of ARGA, improving however the control mechanism by employing a Bayesian decision process. ...
This paper introduces a novel global optimization heuristic algorithm based on the basic paradigms of Evolutionary Algorithms (EA). The algorithm greatly extends a previous strategy proposed by the authors in Munteanu and Lazarescu (1998). In the newly designed algorithm the exploration/exploitation of the search space is adapted on-line based on the current features of the landscape that is being searched. The on-line adaptation mechanism involves a decision process as to whether more exploitation or exploration is needed depending on the current progress of the algorithm and on the current estimated potential of discovering better solutions. The convergence with probability 1 in finite time and discrete space is analyzed, as well as an extensive comparison with other evolutionary optimization heuristics is performed on a set of test functions.
... Our results therefore support a hemispheric asymmetric behavior. This also agrees with the study in [30] where a hidden Markov model and a genetic algorithm were combined to assess hemispheric asymmetry in an MI task. ...
We describe a new technique for the classification of motor imagery electroencephalogram (EEG) recordings in a brain computer interface (BCI) task. The technique is based on an adaptive time-frequency analysis of EEG signals computed using local discriminant bases (LDB) derived from local cosine packets (LCP). In an offline step, the EEG data obtained from the C(3)/C(4) electrode locations of the standard 10/20 system is adaptively segmented in time, over a non-dyadic grid by maximizing the probabilistic distances between expansion coefficients corresponding to left and right hand movement imagery. This is followed by a frequency domain clustering procedure in each adapted time segment to maximize the discrimination power of the resulting time-frequency features. Then, the most discriminant features from the resulting arbitrarily segmented time-frequency plane are sorted. A principal component analysis (PCA) step is applied to reduce the dimensionality of the feature space. This reduced feature set is finally fed to a linear discriminant for classification. The online step simply computes the reduced dimensionality features determined by the offline step and feeds them to the linear discriminant. We provide experimental data to show that the method can adapt to physio-anatomical differences, subject-specific and hemisphere-specific motor imagery patterns. The algorithm was applied to all nine subjects of the BCI Competition 2002. The classification performance of the proposed algorithm varied between 70% and 92.6% across subjects using just two electrodes. The average classification accuracy was 80.6%. For comparison, we also implemented an adaptive autoregressive model based classification procedure that achieved an average error rate of 76.3% on the same subjects, and higher error rates than the proposed approach on each individual subject.
... HMM + SVM Choi 2002, 2003) HMM(Cincotti et al 2003b, Lee and Choi 2003, Liu et al 2003, Obermaier et al 2001a, 2001c, 2001d, Pfurtscheller and Neuper 2001 al 2003) SVM (Blankertz et al 2002a, Guan et al 2004, Garcia et al 2003a, 2003b, 2003c, Garrett et al 2003, Glassman 2005, Gysels and Celka 2004, Gysels et al 2005, Hung et al 2005, Guan et al 2005, Kaper and Ritter 2004a, 2004b, Kaper et al 2004, Lal et al 2004, Peterson et al 2005, Schroder et al 2003, Schröder et al 2005, Thulasidas et al 2004, Trejo et al 2003, Xu et al 2004b, Yom-Tov and Inbar 2001, 2002, 2003, Yoon et al 2005) NID3 (Ivanova et al 1995) CN2 (Ivanova et al 1995) C4.5 (Ivanova et al 1995, Millan et al 2002a) k-NN (Blankertz et al 2002a, Pineda et al 2000, Pregenzer and Pfurtscheller 1999) Threshold detector(Allison andPineda 2003, Balbale et al 1999, Bayliss and Ballard 1999, Bayliss 2003, Bayliss et al 2004, Calhoun and McMillan 1996, Cheng and Gao 1999, Cheng et al 2001, 2005, Donchin et al 2000, Farwell and Donchin 1988, Gao et al 2003b, Graimann et al 2003a, 2003b, 2004, Huggins et al 1999, 2003, Jansen et al 2004, Kawakami et al 1996, Kelly et al 2005b, Kostov and Polak 1997, Lee et al 2005, Levine et al 1999, Levine et al 2000, McMillan and Calhoun 1995, Middendorf et al 2000, Pineda et al 2003, Polak and Kostov 1997, Polikoff et al 1995, Qin et al 2004a He 2005, Roberts et al 1999, Serby et al 2005, Sutter 1992, Wang et al 2004a, 2004b, Xu et al 2004a, Yom-Tov and Inbar 2003) Linear combination − threshold detector (Townsend et al 2004) Continuous feedback + threshold detector ...
Brain-computer interfaces (BCIs) aim at providing a non-muscular channel for sending commands to the external world using the electroencephalographic activity or other electrophysiological measures of the brain function. An essential factor in the successful operation of BCI systems is the methods used to process the brain signals. In the BCI literature, however, there is no comprehensive review of the signal processing techniques used. This work presents the first such comprehensive survey of all BCI designs using electrical signal recordings published prior to January 2006. Detailed results from this survey are presented and discussed. The following key research questions are addressed: (1) what are the key signal processing components of a BCI, (2) what signal processing algorithms have been used in BCIs and (3) which signal processing techniques have received more attention?
... Handedness can be a reason for this asymmetry. It has been recently shown that asymmetric hemisphere modeling can give better results than when perfect mirroring is assumed between both hemispheres [15]. ...
We use local discriminant bases and linear discriminant analysis to classify EEG of left and right hand movement execution and imagination. The local discriminant bases adaptively segment and extract features from real and imagined movement EEG (2003 BCI competition) using cosine packets and Kullback-Leibler, Euclidean and Hellinger class separability (CS) criteria. We also tried principal component analysis (PCA) as another feature reduction method. In our case, CS ordered coefficients resulted in lower classification error than PCA using a smaller number of coefficients. We observed that the most discriminative components were located on the post movement beta and alpha synchronization. Pre-movement features were also selected by the algorithm. We believe that these segments correspond to the mental state and strategy of the subject during the movement execution/imagination. The main advantage of the algorithm is that it adaptively finds these physiological states in an ongoing EEG. This may eliminate the inter- and intra-subject variability. The average error rate of the classification was 12.7% for movement execution and 14.2% for movement imagination. Accordingly, the algorithm would be the 3rd best in the 2003 BCI (brain-computer interface) competition.
Brain Computer Interface (BCI) can be normally defined as the process of controlling the environment with the EEG signals. It is a real-type computer-based process which translates the human brain signals into necessary commands. Because there are many strokes attacked or neurologically affected patients in the world and they are not able to communicate effectively or share their emotions, thoughts with the outside world. Considering some extreme cases, tetraplegic, paraplegic (due to spinal cord injury) or post-stroke patients are factually ‘locked in’ their bodies, incompetent to exert strive motor nerves control after the stroke, paralyses or neurodegenerative diseases, requiring alternative techniques of interactive communication and control of organs. So, BCI is one of the best solutions for this purpose. This paper mainly discusses ensemble learning approaches for EEG signal classification and feature extraction. Bagging, Adaptive boosting, and Gradient boosting are quite popular ensemble learning methods, which are very effective for elucidation and explication of many practical classification problems. EEG signals may be classified as adopting a set of features like autoregression, power spectrum density, energy entropy, and linear complexity. In this paper, we have used three illustrative procedures Adaptive Boosting, Bagging, and Gradient Boosting ensemble learning methods and extracted features using discrete wavelet transform (DWT), autoregression (AR), power spectrum density (PSD), and common spatial pattern (CSP).
This PhD thesis consists in jointly analyzing eye-tracking signals and multi-channel electroencephalograms (EEGs) acquired concomitantly on participants doing an information collection reading task in order to take a binary decision - is the text related to some topic or not ? Textual information search is not a homogeneous process in time - neither on a cognitive point of view, nor in terms of eye-movement. On the contrary, this process involves several steps or phases, such as normal reading, scanning, careful reading - in terms of oculometry - and creation and rejection of hypotheses, confirmation and decision - in cognitive terms.In a first contribution, we discuss an analysis method based on hidden semi-Markov chains on the eye-tracking signals in order to highlight four interpretable phases in terms of information acquisition strategy: normal reading, fast reading, careful reading, and decision making.In a second contribution, we link these phases with characteristic changes of both EEGs signals and textual information. By using a wavelet representation of EEGs, this analysis reveals variance and correlation changes of the inter-channels coefficients, according to the phases and the bandwidth. And by using word embedding methods, we link the evolution of semantic similarity to the topic throughout the text with strategy changes.In a third contribution, we present a new model where EEGs are directly integrated as output variables in order to reduce the state uncertainty. This novel approach also takes into consideration the asynchronous and heterogeneous aspects of the data.
During the past 20 years, scientists have focused their efforts in the quest of real solutions in which the neural signals produced inside the human brain could be connected with computers or artificial prostheses that in a near future could be used to restore the mobility and communication abilities of patients with some damage in the central nervous system. In this paper, the procedure to control an artificial device with the thought is explained; the techniques used to extract the neural activity from the brain are classified and compared to establish their advantages, drawbacks, and future development. In addition, the main breakthroughs so far in brain-machine interfaces are described.
This contribution presents our advances in development of a real-time EEG processing system. Existing system was extended by common configuration, logging system and new modules including support for both new EEG device and new experiments. Correct function of the system was extensively tested and verified in numerous experiments with several subjects and demonstrated at the faculty open day. Scores of 75% were achieved with classification of left and right hand movement imagery. We use the simplest possible methods as we focus on feedback influence. Difficulties of transition to a real BCI operation along with faced problems are discussed.
This paper addresses the question of maximizing classifier accuracy for classifying task-related mental activity from Magnetoencelophalography (MEG) data. We propose the use of different sources of information and introduce an automatic channel selection procedure. To determine an informative set of channels, our approach combines a variety of machine learning algorithms: feature subset selection methods, classifiers based on regularized logistic regression, information fusion, and multiobjective optimization based on probabilistic modeling of the search space. The experimental results show that our proposal is able to improve classification accuracy compared to approaches whose classifiers use only one type of MEG information or for which the set of channels is fixed a priori.
A brain computer interface (BCI) transforms electrophysiological signals originating from the human brain into commands that control devices or applications. In this way, a BCI provides a new nonmuscular communication channel, which can be extremely useful for people with severe neuromuscular disorders such as amyotrophic lateral sclerosis and brainstem stroke. The immediate goal of our current research is to provide these users with an opportunity to communicate with their environment. Future applications also include the field of multimedia and virtual reality.
Present day BCIs use a variety of electrophysiological signals such as slow cortical potentials, evoked potentials (P300), oscillatory activity recorded from scalp or subdural electrodes, and cortical neuronal activity recorded from implanted electrodes. EEG is by far the most frequently used input source, because it is readily available and noninvasive. The poor signal‐to‐noise ratio of scalp–recorded signals requires the application of advanced signal processing methods.
This article outlines and explains the current approaches and methods used in BCI research with emphasis on the signal processing part of the system, consisting of preprocessing, feature extraction, and classification. However, the success of a BCI depends not only on the methodologies applied, but also on the capability of the user to develop and maintain the skill to produce the brain patterns employed by the BCI. Therefore, the interaction between the BCI system and the user, in terms of adaptation and learning, is a challenging aspect of any BCI development and application.
It is now common knowledge that blind search algorithms cannot perform with equal efficiency on all possible optimization
problems defined on a domain. This knowledge applies also to Genetic Algorithms when viewed as global and blind optimizers.
From this point of view it is necessary to design algorithms capable of adapting their search behavior by making use in a
direct fashion of the knowledge pertaining to the search landscape. The paper introduces a novel adaptive Genetic Algorithm
where the exploration / exploitation is directly controlled during evolution using a Bayesian decision process. Test cases
are analyzed as to how parameters affect the search behavior of the algorithm.
Bagging, boosting and random subspace are three popular ensemble learning methods, which have already shown effectiveness
in many practical classification problems. For electroencephalogram (EEG) signal classification arising in recent brain-computer
interface (BCI) research, however, there are almost no reports investigating their feasibilities. This paper systematically
evaluates the performance of these three ensemble methods for their new application on EEG signal classification. Experiments
are conducted on three BCI subjects with k-nearest neighbor and decision tree as base classifiers. Several valuable conclusions
are derived about the feasibility and performance of ensemble methods for classifying EEG signals.
Self-paced Brain computer interface (SBCI) systems allow individuals with motor disabilities to use their brain signals to control devices, whenever they wish. These systems are required to identify the user’s “intentional control (IC)” commands and they must remain inactive during all periods in which users do not intend control (called “no control (NC)” periods). This dissertation addresses three issues related to the design of SBCI systems: 1) their presently high false positive (FP) rates, 2) the presence of artifacts and 3) the identification of a suitable evaluation metric. To improve the performance of SBCI systems, the following are proposed: 1) a method for the automatic user-customization of a 2-state SBCI system, 2) a two-stage feature reduction method for selecting wavelet coefficients extracted from movement-related potentials (MRP), 3) an SBCI system that classifies features extracted from three neurological phenomena: MRPs, changes in the power of the Mu and Beta rhythms; 4) a novel method that effectively combines methods developed in 2) and 3 ) and 5) generalizing the system developed in 3) for detecting a right index finger flexion to detecting the right hand extension. Results of these studies using actual movements show an average true positive (TP) rate of 56.2% at the FP rate of 0.14% for the finger flexion study and an average TP rate of 33.4% at the FP rate of 0.12% for the hand extension study. These FP results are significantly lower than those achieved in other SBCI systems, where FP rates vary between 1-10%. We also conduct a comprehensive survey of the BCI literature. We demonstrate that many BCI papers do not properly deal with artifacts. We show that the proposed BCI achieves a good performance of TP=51.8% and FP=0.4% in the presence of eye movement artifacts. Further tests of the performance of the proposed system in a pseudo-online environment, shows an average TP rate =48.8% at the FP rate of 0.8%. Finally, we propose a framework for choosing a suitable evaluation metric for SBCI systems. This framework shows that Kappa coefficient is more suitable than other metrics in evaluating the performance during the model selection procedure.
In this work we present the first comprehensive survey of Brain Interface (BI) technology designs published prior to January 2006. Detailed results from this survey, which was based on the Brain Interface Design Framework proposed by Mason and Birch, are presented and discussed to address the following research questions: (1) which BI technologies are directly comparable, (2) what technology designs exist, (3) which application areas (users, activities and environments) have been targeted in these designs, (4) which design approaches have received little or no research and are possible opportunities for new technology, and (5) how well are designs reported. The results of this work demonstrate that meta-analysis of high-level BI design attributes is possible and informative. The survey also produced a valuable, historical cross-reference where BI technology designers can identify what types of technology have been proposed and by whom.
It is widely accepted in the brain computer interface (BCI) research community that neurological phenomena are the only source of control in any BCI system. Artifacts are undesirable signals that can interfere with neurological phenomena. They may change the characteristics of neurological phenomena or even be mistakenly used as the source of control in BCI systems. Electrooculography (EOG) and electromyography (EMG) artifacts are considered among the most important sources of physiological artifacts in BCI systems. Currently, however, there is no comprehensive review of EMG and EOG artifacts in BCI literature. This paper reviews EOG and EMG artifacts associated with BCI systems and the current methods for dealing with them. More than 250 refereed journal and conference papers are reviewed and categorized based on the type of neurological phenomenon used and the methods employed for handling EOG and EMG artifacts. This study reveals weaknesses in BCI studies related to reporting the methods of handling EMG and EOG artifacts. Most BCI papers do not report whether or not they have considered the presence of EMG and EOG artifacts in the brain signals. Only a small percentage of BCI papers report automated methods for rejection or removal of artifacts in their systems. As the lack of dealing with artifacts may result in the deterioration of the performance of a particular BCI system during practical applications, it is necessary to develop automatic methods to handle artifacts or to design BCI systems whose performance is robust to the presence of artifacts.
EEG recordings provide an important means of brain-computer communication, but their classification accuracy is limited by unforeseeable variations in the signal due to artefacts or recogniser-subject feedback. A number of techniques were recently developed to address a related problem of recogniser robustness to uncontrollable signal variation which also occurs in automatic speech recognition (ASR). In this article we consider how some of the proved advantages of the "multi-stream combination" and "tandem" approaches in HMM/ANN hybrid based ASR can possibly be applied to improve the performance of EEG recognition. Keywords: EEG, multi-stream classification, robust recognition Acknowledgements: This work was supported by the EC/OFES (European Community / Swiss Federal Office for Education and Science) RESPITE project (REcognition of Speech by Partial Information TEchniques). Contents 1.
Hidden Markov models (HMMs) are presented for the online classification of single trial EEG data during imagination of a left or right hand movement. The classification shows an improvement of the online experiment and the temporal determination of minimal classification error compared to linear classification methods.
In this paper we study random genetic drift in a finite genetic population. Exact formulae for calculating the mean convergence time of the population are analytically derived and some results of numerical calculations are given. The calculations are compared to the results obtained in population genetics. A new proposition is derived for binary alleles and uniform crossover. Here the mean convergence time is almost proportional to the size of the population and to the logarithm of the number of the loci. The results of Monte Carlo type numerical simulations are in agreement with the results from the calculation.
Electroencephalogram (EEG) recordings during right and left motor imagery can be used to move a cursor to a target on a computer screen. Such an EEG-based brain-computer interface (BCI) can provide a new communication channel to replace an impaired motor function. It can be used by, e.g., patients with amyotrophic lateral sclerosis (ALS) to develop a simple binary response in order to reply to specific questions. Four subjects participated in a series of on-line sessions with an EEG-based cursor control. The EEG was recorded from electrodes overlying sensory-motor areas during left and right motor imagery. The EEG signals were analyzed in subject-specific frequency bands and classified on-line by a neural network. The network output was used as a feedback signal. The on-line error (100%-perfect classification) was between 10.0 and 38.1%. In addition, the single-trial data were also analyzed off-line by using an adaptive autoregressive (AAR) model of order 6. With a linear discriminant analysis the estimated parameters for left and right motor imagery were separated. The error rate obtained varied between 5.8 and 32.8% and was, on average, better than the on-line results. By using the AAR-model for on-line classification an improvement in the error rate can be expected, however, with a classification delay around 1 s.
When Jean-Dominique Bauby suffered from a cortico-subcortical stroke that led to complete paralysis with totally intact sensory and cognitive functions, he described his experience in The Diving-Bell and the Butterfly as ``something like a giant invisible diving-bell holds my whole body prisoner''. This horrifying condition also occurs as a consequence of a progressive neurological disease, amyotrophic lateral sclerosis, which involves progressive degeneration of all the motor neurons of the somatic motor system. These `locked-in' patients ultimately become unable to express themselves and to communicate even their most basic wishes or desires, as they can no longer control their muscles to activate communication devices. We have developed a new means of communication for the completely paralysed that uses slow cortical potentials (SCPs) of the electro-encephalogram to drive an electronic spelling device.
Over the past decade, many laboratories have begun to explore brain-computer interface (BCI) technology as a radically new communication option for those with neuromuscular impairments that prevent them from using conventional augmentative communication methods. BCI's provide these users with communication channels that do not depend on peripheral nerves and muscles. This article summarizes the first international meeting devoted to BCI research and development. Current BCI's use electroencephalographic (EEG) activity recorded at the scalp or single-unit activity recorded from within cortex to control cursor movement, select letters or icons, or operate a neuroprosthesis. The central element in each BCI is a translation algorithm that converts electrophysiological input from the user into output that controls external devices. BCI operation depends on effective interaction between two adaptive controllers, the user who encodes his or her commands in the electrophysiological input provided to the BCI, and the BCI which recognizes the commands contained in the input and expresses them in device control. Current BCI's have maximum information transfer rates of 5-25 b/min. Achievement of greater speed and accuracy depends on improvements in signal processing, translation algorithms, and user training. These improvements depend on increased interdisciplinary cooperation between neuroscientists, engineers, computer programmers, psychologists, and rehabilitation specialists, and on adoption and widespread application of objective methods for evaluating alternative methods. The practical use of BCI technology depends on the development of appropriate applications, identification of appropriate user groups, and careful attention to the needs and desires of individual users. BCI research and development will also benefit from greater emphasis on peer-reviewed publications, and from adoption of standard venues for presentations and discussion.
The development of an electroencephalograph (EEG)-based brain-computer interface (BCI) requires rapid and reliable discrimination of EEG patterns, e.g., associated with imaginary movement. One-sided hand movement imagination results in EEG changes located at contra- and ipsilateral central areas. We demonstrate that spatial filters for multichannel EEG effectively extract discriminatory information from two populations of single-trial EEG, recorded during left- and right-hand movement imagery. The best classification results for three subjects are 90.8%, 92.7%, and 99.7%. The spatial filters are estimated from a set of data by the method of common spatial patterns and reflect the specific activation of cortical areas. The method performs a weighting of the electrodes according to their importance for the classification task. The high recognition rates and computational simplicity make it a promising method for an EEG-based brain-computer interface.
The electroencephalogram (EEG) is modified by motor imagery and can be used by patients with severe motor impairments (e.g., late stage of amyotrophic lateral sclerosis) to communicate with their environment. Such a direct connection between the brain and the computer is known as an EEG-based brain-computer interface (BCI). This paper describes a new type of BCI system that uses rapid prototyping to enable a fast transition of various types of parameter estimation and classification algorithms to real-time implementation and testing. Rapid prototyping is possible by using Matlab, Simulink, and the Real-Time Workshop. It is shown how to automate real-time experiments and perform the interplay between on-line experiments and offline analysis. The system is able to process multiple EEG channels on-line and operates under Windows 95 in real-time on a standard PC without an additional digital signal processor (DSP) board. The BCI can be controlled over the Internet, LAN or modem. This BCI was tested on 3 subjects whose task it was to imagine either left or right hand movement. A classification accuracy between 70% and 95% could be achieved with two EEG channels after some sessions with feedback using an adaptive autoregressive (AAR) model and linear discriminant analysis (LDA).
The information transfer rate, given in bits per trial, is used as an evaluation measurement in a brain-computer interface (BCI). Three subjects performed four motor-imagery (left hand, right hand, foot, and tongue) and one mental-calculation task. Classification of the electroencephalogram (EEG) patterns is based on band power estimates and hidden Markov models (HMMs). We propose a method that combines the EEG patterns based on separability into subsets of two, three, four, and five mental tasks. The information transfer rates of the BCI systems comprised of these subsets are reported. The achieved information transfer rates vary from 0.42 to 0.81 bits per trial and reveal that the upper limit of different mental tasks for a BCI system is three. In each subject, different combinations of three tasks resulted in the best performance.
This article explores the use of scalar and multivariate autoregressive (AR) models to extract features from the human electroencephalogram (EEG) with which mental tasks can be discriminated. This is part of a larger project to investigate the feasibility of using EEG to allow paralyzed persons to control a device such as a wheelchair. EEG signals from four subjects were recorded while they performed two mental tasks. Quarter-second windows of six-channel EEG were transformed into four different representations: scalar AR model coefficients, multivariate AR coefficients, eigenvalues of a correlation matrix, and the Karhunen-Loeve transform of the multivariate AR coefficients. Feature vectors defined by these representations were classified with a standard, feedforward neural network trained via the error backpropagation algorithm. The four representations produced similar results, with the multivariate AR coefficients performing slightly better and more consistently with an average classification accuracy of 91.4% on novel, untrained, EEG signals.
Hidden Markov models (HMM) are introduced for the offline classification of single-trail EEG data in a brain-computer interface (BCI). The HMMs are used to classify Hjorth parameters calculated from bipolar EEG data, recorded during the imagination of a left or right hand movement. The effects of different types of HMMs on the recognition rate are discussed. Furthermore a comparison of the results achieved with the linear discriminat (LD) and the HMM, is presented.
In this paper we introduce a new search heuristic based on a Genetic Algorithm structure, called Adaptive Reservoir Genetic Algorithm (ARGA). This search mechanism reduces the loss of genetic diversity, present in most of Genetic Algorithms, by employing an adaptive mutant subpopulation called reservoir. We present preliminary theoretical considerations and heuristics that support ARGA. These considerations, as well as the experimental results obtained on three optimisation test functions, show that ARGA is an effective optimisation heuristic, less affected by premature convergence and genetic drift, that are common problems in many Genetic Algorithms.
Hidden Markov models (HMMs) have during the last decade become a widely spread tool for modelling sequences of dependent random variables. Inference for HMMs has been considered by several authors, but so far no work has been done on estimating their order. In this paper we propose a penalized likelihood estimator for this purpose. This estimator is based on the m-dimensional distribution of HMM, and it is shown that in the limit it does not underestimate the order.
Scitation is the online home of leading journals and conference proceedings from AIP Publishing and AIP Member Societies
S ummary
A broadly applicable algorithm for computing maximum likelihood estimates from incomplete data is presented at various levels of generality. Theory showing the monotone behaviour of the likelihood and convergence of the algorithm is derived. Many examples are sketched, including missing value situations, applications to grouped, censored or truncated data, finite mixture models, variance component estimation, hyperparameter estimation, iteratively reweighted least squares and factor analysis.
A comparison is made between the dynamics of steady state and generational genetic algorithms using the statistical mechanics approach developed by Prugel-Bennett, Shapiro and Rattray. It is shown that the loss of variance of the population under steady state selection - genetic drift - occurs at twice the rate of generational selection. By considering a simple ones counting problem with selection and mutation, it is shown that, with weak selection, the steady state genetic algorithm can reproduce the dynamics of the generational genetic algorithm at half the computational cost in terms of function evaluations.
Genetic algorithms are playing an increasingly important role in studies of complex adaptive systems, ranging from adaptive agents in economic theory to the use of machine learning techniques in the design of complex devices such as aircraft turbines and integrated circuits. Adaptation in Natural and Artificial Systems is the book that initiated this field of study, presenting the theoretical foundations and exploring applications.
In its most familiar form, adaptation is a biological process, whereby organisms evolve by rearranging genetic material to survive in environments confronting them. In this now classic work, Holland presents a mathematical model that allows for the nonlinearity of such complex interactions. He demonstrates the model's universality by applying it to economics, physiological psychology, game theory, and artificial intelligence and then outlines the way in which this approach modifies the traditional views of mathematical genetics.
Initially applying his concepts to simply defined artificial systems with limited numbers of parameters, Holland goes on to explore their use in the study of a wide range of complex, naturally occuring processes, concentrating on systems having multiple factors that interact in nonlinear ways. Along the way he accounts for major effects of coadaptation and coevolution: the emergence of building blocks, or schemata, that are recombined and passed on to succeeding generations to provide, innovations and improvements.
Bradford Books imprint
A new communication channel for severely handicapped people could be opened with a direct brain to computer interface (BCI). Such a system classifies electrical brain signals online. In a series of training sessions, where electroencephalograph (EEG) signals are recorded on the intact scalp, a classifier is trained to discriminate a limited number of different brain states. In a subsequent series of feedback sessions, where the subject is confronted with the classification results, the subject tries to reduce the number of misclassifications. In this study the relevance of different spectral components is analyzed: 1) on the training sessions to select optimal frequency bands for the feedback sessions and 2) on the feedback sessions to monitor changes.
Thesis (M.S.)--Colorado State University, 1995. Includes bibliographical references.
A method to describe the general characteristics of an EEG trace in a few quantitative terms is introduced. Its descriptive parameters are entirely based on time, but they can be derived also from the statistical moments of the power spectrum. Thus the method provides a bridge between a physical time domain interpretation and the conventional frequency domain description. Further, the parameters are based on the concept of variance, giving them an additive property so that the measured values pertain also to any basic elements from which a complex curve may be composed by superposition.The proposed method offers a way to on-line measurement of basic signal properties by means of a time-based calculation, requiring less complex equipment compared to conventional frequency analysis. The data-reducing capability of the parameters has been experimentally stated in the recording of “sleep profiles”.RésuméL'auteur introduit une méthode de description des caractéristiques générales d'un tracé EEG en un nombre limité de termes quantitatifs. Ses paramètres descriptifs sont entièrement basés sur le temps, mais peuvent être dérivés également des moments statistiques du spectre de puissance. Ainsi, cette méthode fait la jonction entre une interprétation du domaine des séries temporelles physiques et la description du domaine fréquentiel conventionnel. De plus les paramètres sont basés sur le concept de variance, leur donnant une propriété supplémentaire de telle sorte que les valeurs mesurées se rapportent également à chaque élément de base à partir duquel une courbe complexe peut être composée par superposition.La méthode proposée offre un moyen de mesurer “on-line” des propriétés de base du signal au moyen d'un calcul basé sur le temps, nécessitant un équipement moins complexe que l'analyse de fréquence conventionnelle. La capacité de réduction des données des paramètries a été spécifiée expérimentalement dans l'enregistrement des “profils de sommeil”.
One major question in designing an EEG-based Brain Computer Interface to bypass the normal motor pathways is the selection of proper electrode positions. This study investigates electrode selection with a Distinction Sensitive Learning Vector Quantizer (DSLVQ). DSLVQ is an extended Learning Vector Quantizer (LVQ) which employs a weighted distance function for dynamical scaling and feature selection. The data analysed and classified were 56-channel EEG recordings over sensorimotor areas during preparation for discrete left or right index finger flexions. Data from 3 subjects are reported. It was found by DSLVQ that the most important electrode positions for differentiation between planning of left and right finger movement overlie cortical finger/hand areas over both hemispheres.
Hidden Markov models (HMM) are introduced for the offline classification of single-trail EEG data in a brain-computer-interface (BCI). The HMMs are used to classify Hjorth parameters calculated from bipolar EEG data, recorded during the imagination of a left or right hand movement. The effects of different types of HMMs on the recognition rate are discussed. Furthermore a comparison of the results achieved with the linear discriminant (LD) and the HMM, is presented.
Stimulus-related changes in ongoing electroencephalography (EEG) over sensorimotor areas were investigated during a visually cued motor imagery task. Four subjects were instructed to imagine one-sided hand movements in response to visual cue stimuli. The EEG was recorded from central areas using 27 electrodes set at distances of 2.5 cm. The method of common spatial filters was used to extract discriminatory information of EEG patterns recorded during the two motor imagery conditions. Single EEG trials were classified in intervals of 250 ms for a 8-s period starting 3 s prior to stimulus presentation. The results suggest that perception of the visual cue stimulus modifies oscillations in sensorimotor areas specific to the indicated hand starting as soon as 250-500 ms after stimulus onset.
EEG feedback studies demonstrate that human subjects can learn to regulate electrocortical activity over the sensorimotor cortex. Such self-induced EEG changes could serve as control signals for a Brain Computer Interface. The experimental task of the current study was to imagine either right-hand or left-hand movement depending on a visual cue stimulus on a computer monitor. The performance of this imagination task was controlled on-line by means of a feedback bar that represented the current EEG pattern. EEG signals recorded from left and right central recording sites were used for on-line classification. For the estimation of EEG parameters, an adaptive autoregressive model was applied, and a linear discriminant classifier was used to discriminate between EEG patterns associated with left and right motor imagery. Four trained subjects reached 85% to 95% classification accuracy in the course of the experimental sessions. To investigate the impact of continuous feedback presentation, time courses of band power changes were computed for subject-specific frequency bands. The EEG data revealed a significant event-related desynchronization over the contralateral central area in all subjects. Two subjects simultaneously displayed synchronization of EEG activity (event-related synchronization) over the ipsilateral side. During feedback presentation the event-related desynchronization/event-related synchronization patterns showed increased hemispheric asymmetry compared to initial control sessions without feedback.
This paper describes a research approach to develop a brain-computer interface (BCI) based on recognition of subject-specific EEG patterns. EEG signals recorded from sensorimotor areas during mental imagination of specific movements are classified on-line and used e.g. for cursor control. In a number of on-line experiments, various methods for EEG feature extraction and classification have been evaluated.
This paper introduces genetic algorithms (GA) as a complete
entity, in which knowledge of this emerging technology can be integrated
together to form the framework of a design tool for industrial
engineers. An attempt has also been made to explain “why”
and “when” GA should be used as an optimization tool
There has been a lot of recent interest in so-called "steady state" genetic algorithms (GAs) which, among other things, replace only a few individuals (typically 1 or 2) each generation from a fixed size population of size N. Understanding the advantages and/or disadvantages of replacing only a fraction of the population each generation (rather than the entire population) was a goal of some of the earliest GA research. In spite of considerable progress in our understanding of GAs since then, the pros/cons of overlapping generations remains a somewhat cloudy issue. However, recent theoretical and empirical results provide the background for a much clearer understanding of this issue. In this paper we review, combine, and extend these results in a way that significantly sharpens our insight. 1 INTRODUCTION In Holland's book Adaptation in Natural and Artificial Systems [Holland75], he introduces and analyzes two forms of reproductive plans which have served as the basis for the field of ...
A comparison is made between the dynamics of steady state and generational genetic algorithms using the statistical mechanics approach developed by PrugelBennett, Shapiro and Rattray. It is shown that the loss of variance of the population under steady state selection --- genetic drift --- occurs at twice the rate of generational selection. By considering a simple ones counting problem with selection and mutation, it is shown that, with weak selection, the steady state genetic algorithm can reproduce the dynamics of the generational genetic algorithm at half the computational cost in terms of function evaluations. 1 Introduction Since the popularisation of the genetic algorithm (GA) by Holland [1], there have been two strategies for reproducing the population members. In one, the entire population is replaced simultaneously --- the generational GA. In the steady state GA, populations overlap. One or two population members are reproduced in each time frame. The study of these two schem...
EEG event-related desynchronization (ERD) and event-related synchronization (ERS), " in Electroencephalog-raphy, Basic Principles, Clinical Applications and Related Fields
- G Pfurtscheller
- Systems
- And Man
- Cybernetics—part
G. Pfurtscheller, " EEG event-related desynchronization (ERD) and event-related synchronization (ERS), " in Electroencephalog-raphy, Basic Principles, Clinical Applications and Related Fields. Amsterdam, The Netherlands: Elsevier, 1998. IEEE TRANSACTIONS ON SYSTEMS, MAN, AND CYBERNETICS—PART C: APPLICATIONS AND REVIEWS, VOL. 31, NO. 4, NOVEMBER 2001
Ahn are with Mechatronics, Kwang-Ju Institute of
- S.-C Byun
S.-C. Byun and B.-H. Ahn are with Mechatronics, Kwang-Ju Institute of Science and Technology (K-JIST), Kwangju 500-712, Korea (e-mail: scbyun@kjist.ac.kr; bayhay@kjist.ac.kr).
Motor imagery and ERD
- G Pfurtscheller
- F H L Da Silva
Generations gaps revisited
- DeJong