Objective: Most current Electroencephalography (EEG)-based Brain-Computer Interfaces (BCIs) are based on machine learning algorithms. There is a large diversity of classifier types that are used in this field, as described in our 2007 review paper. Now, approximately 10 years after this review publication, many new algorithms have been developed and tested to classify EEG signals in BCIs. The time is therefore ripe for an updated review of EEG classification algorithms for BCIs. Approach: We surveyed the BCI and machine learning literature from 2007 to 2017 to identify the new classification approaches that have been investigated to design BCIs. We synthesize these studies in order to present such algorithms, to report how they were used for BCIs, what were the outcomes, and to identify their pros and cons. Main results: We found that the recently designed classification algorithms for EEG-based BCIs can be divided into four main categories: adaptive classifiers, matrix and tensor classifiers, transfer learning and deep learning, plus a few other miscellaneous classifiers. Among these, adaptive classifiers were demonstrated to be generally superior to static ones, even with unsupervised adaptation. Transfer learning can also prove useful although the benefits of transfer learning remain unpredictable. Riemannian geometry-based methods have reached state-of-the-art performances on multiple BCI problems and deserve to be explored more thoroughly, along with tensor-based methods. Shrinkage linear discriminant analysis and random forests also appear particularly useful for small training samples settings. On the other hand, deep learning methods have not yet shown convincing improvement over state-of-the-art BCI methods. Significance: This paper provides a comprehensive overview of the modern classification algorithms used in EEG-based BCIs, presents the principles of these Review of Classification Algorithms for EEG-based BCI 2 methods and guidelines on when and how to use them. It also identifies a number of challenges to further advance EEG classification in BCI.

https://iopscience.iop.org/article/10.1088/1741-2552/aab2f2/pdf

A Review of Classification Algorithms for EEG-based Brain-Computer Interfaces: A 10-year Update

Common spatial pattern (CSP) has been an effective technique for feature extraction in electroencephalography (EEG) based brain computer interfaces (BCIs). However, motor imagery EEG signal feature extraction using CSP generally depends on the selection of the frequency bands to a great extent. In this study, we propose a mutual information based frequency band selection approach. The idea of the proposed method is to utilize the information from all the available channels for effectively selecting the most discriminative filter banks. CSP features are extracted from multiple overlapping sub-bands. An additional sub-band has been introduced that cover the wide frequency band (7–30 Hz) and two different types of features are extracted using CSP and common spatio-spectral pattern techniques, respectively. Mutual information is then computed from the extracted features of each of these bands and the top filter banks are selected for further processing. Linear discriminant analysis is applied to the features extracted from each of the filter banks. The scores are fused together, and classification is done using support vector machine. The proposed method is evaluated using BCI Competition III dataset IVa, BCI Competition IV dataset I and BCI Competition IV dataset IIb, and it outperformed all other competing methods achieving the lowest misclassification rate and the highest kappa coefficient on all three datasets. Introducing a wide sub-band and using mutual information for selecting the most discriminative sub-bands, the proposed method shows improvement in motor imagery EEG signal classification.

/pdf/an-improved-discriminative-filter-bank-selection-approach-2no47vovvj.pdf

An improved discriminative filter bank selection approach for motor imagery EEG signal classification using mutual information

Reinforcement learning (RL) enables robots to learn its optimal behavioral strategy in dynamic environments based on feedback. Explicit human feedback during robot RL is advantageous, since an explicit reward function can be easily adapted. However, it is very demanding and tiresome for a human to continuously and explicitly generate feedback. Therefore, the development of implicit approaches is of high relevance. In this paper, we used an error-related potential (ErrP), an event-related activity in the human electroencephalogram (EEG), as an intrinsically generated implicit feedback (rewards) for RL. Initially we validated our approach with seven subjects in a simulated robot learning scenario. ErrPs were detected online in single trial with a balanced accuracy (bACC) of 91%, which was sufficient to learn to recognize gestures and the correct mapping between human gestures and robot actions in parallel. Finally, we validated our approach in a real robot scenario, in which seven subjects freely chose gestures and the real robot correctly learned the mapping between gestures and actions (ErrP detection (90% bACC)). In this paper, we demonstrated that intrinsically generated EEG-based human feedback in RL can successfully be used to implicitly improve gesture-based robot control during human-robot interaction. We call our approach intrinsic interactive RL.

/pdf/intrinsic-interactive-reinforcement-learning-using-error-2nfudiz6km.pdf

Intrinsic interactive reinforcement learning – Using error-related potentials for real world human-robot interaction

An iterative wavelet threshold for signal denoising

The recent revolutions in Internet of Things (IoT) and big data analytics have unlocked promising possibilities in biomedical and health care technologies. The current chapter deals with theoretical, methodological, validated, and empirical concepts with exciting examples related to the subject. First, the application of IoT and big data in the analysis of a vast image database generated every day from various sources by big data combined with machine-learning algorithms and other forms of artificial intelligence, to generate structured information for application in remote diagnostics, is explained with examples. Further emerging trends of telemedicine using artificial intelligence technique in robotics health care have been discussed. The flexibility and adaptability of AI-based telemedicine provide the endless supports for health care development can be observed in the literature. The advancement of telerobotic surgery and the importance of the Internet of Robotic Things (IoRT) have been discussed briefly. Furthermore, details on wearable devices already available targeting the biomedical and health care applications that are capable of collecting, analyzing by standard protocols by machine intelligence for predictions of health-related issues are described. Lastly, the final expansion and evolution of these technologies in the modern health care system and biomedical research, starting with personalized drug designing to targeted drug delivery and beyond it, is described.

Emerging trends in IoT and big data analytics for biomedical and health care technologies

Goal: Current brain–computer interfaces (BCIs) are usually based on various, often supervised, signal processing methods. The disadvantage of supervised methods is the requirement to calibrate them with recently acquired subject-specific training data. Here, we present a novel algorithm for dimensionality reduction (spatial filter), that is ideally suited for single-trial detection of event-related potentials (ERPs) and can be adapted online to a new subject to minimize or avoid calibration time. Methods: The algorithm is based on the well-known xDAWN filter, but uses generalized eigendecomposition to allow an incremental training by recursive least squares (RLS) updates of the filter coefficients. We analyze the effectiveness of the spatial filter in different transfer scenarios and combinations with adaptive classifiers. Results: The results show that it can compensate changes due to switching between different users, and therefore allows to reuse training data that has been previously recorded from other subjects. Conclusions: The presented approach allows to reduce or completely avoid a calibration phase and to instantly use the BCI system with only a minor decrease of performance. Significance: The novel filter can adapt a precomputed spatial filter to a new subject and make a BCI system user independent.

An Adaptive Spatial Filter for User-Independent Single Trial Detection of Event-Related Potentials

In today’s space applications, astronauts have to perform a variety of tasks. An intelligent on-board monitoring system that ensures uninterrupted, on time monitoring of the astronauts would be a great support. In this paper we want to introduce a new approach for the use of single trial analysis of the human electroencephalogram (EEG) that gives insight into the cognitive state of the operator. This passive monitoring is called brain reading (BR). We discuss the suitability of BR for operator support in space applications regarding the achieved kind and effectiveness of assistance. Further on, solutions for method-depending constraints like the acquisition of training data as well as the the need for software and hardware improvements that will allow the integration of BR in real space applications in the future, are explained.

/pdf/towards-operator-monitoring-via-brain-reading-an-eeg-based-1lsbs4x3o7.pdf

Towards Operator Monitoring via Brain Reading - An EEG-based Approach for Space Applications

Current movement prediction systems based on electroencephalography were mainly developed and evaluated in highly controlled scenarios, in which subjects concentrate only on the desired task with as few as possible disturbing sources present. However, it has not been addressed sufficiently how the suggested methods perform in more complex and uncontrolled environments. In this work we predict arm movements online in a robotic teleoperation scenario and present a completely online running methodology. The system is evaluated on ten sessions from three subjects. Evaluation criteria are the overall classification performance and the success in predicting an upcoming movement in the application. Our results confirm that it is possible to predict movements in less restricted applications motivating the transfer of these methods to real world applications.

Online movement prediction in a robotic application scenario

A Framework for High Performance Embedded Signal Processing and Classification of Psychophysiological Data

The detection of event related potentials and their usage for innovative applications became an increasingly important research topic for brain computer interfaces in the last couple of years. However, brain computer interfaces use methods that need to be trained on subject-specific data before they can be used. This problem must be solved for real-world applications in which humans are multi tasking and hence are to some degree are less predictable in their behavior compared to classical set ups for brain computer interfacing. In this paper, we show the detection and passive usage of the P300 related brain activity in a highly uncontrolled and noisy application scenario. The subjects are multi tasking, i.e., they perform a demanding senso-motor task, i.e., the telemanipulate a real robotic arm while responding to important messages. For telemanipulation, the subject wears an active exoskeleton to control a robotic arm, which is presented to him in a virtual scenario. By online analysis of the subject’s electroencephalogram we detect P300 related target recognition processes to infer on upcoming response behavior on presented task-relevant messages (Targets) or missing of response behavior in case a Target was not recognized. We show that a classifier that is trained to distinguish between brain activity evoked by recognized task-relevant stimuli (recognized Targets) and ignored frequent task-irrelevant stimuli (Standards) can be applied to classify between brain activity evoked by recognized targets and brain activity that is evoked in case that task-relevant stimuli are not recognized (Missed Targets). The applied transfer of classifier results in reduced performance. We show that this draw back of the approach can strongly be improved by using online machine learning tools to adapt the pre-trained classifier to the new class, i.e., to the Missed Target class, that was not used during training of the classifier.

Hendrik Woehrle

Papers

An Adaptive Spatial Filter for User-Independent Single Trial Detection of Event-Related Potentials

Towards Operator Monitoring via Brain Reading - An EEG-based Approach for Space Applications

Online movement prediction in a robotic application scenario

A Framework for High Performance Embedded Signal Processing and Classification of Psychophysiological Data

Online Classifier Adaptation for the Detection of P300 Target Recognition Processes in a Complex Teleoperation Scenario