A review of classification algorithms for EEG-based brain–computer interfaces: a 10 year update

Lotte, Fabien; Bougrain, Laurent; Cichocki, Andrzej; Clerc, Maureen; Congedo, Marco; Rakotomamonjy, Alain; Yger, Florian

doi:10.1088/1741-2552/aab2f2

Cited by 1,439 publications

(1,006 citation statements)

References 236 publications

Supporting

Mentioning

982

Contrasting

Unclassified

Order By: Relevance

“…This includes intersubject variability, uncertain long-term effects, and the apparent failure of some individuals to achieve self-regulation (6). To tackle these issues, investigators have searched for better decoders of neural activity (7) as well as for psychological factors (8) and appropriate training regimens (9) that can influence the user's performance. On the other hand, neuroplasticity is thought to be crucial for achieving effective control and this has motivated a deeper understanding of the neurophysiological mechanisms of neurofeedback and BCI learning (10).…”

Section: Introductionmentioning

confidence: 99%

Spatiotemporal neural correlates of brain-computer interface learning

Corsi

Chávez

Schwartz

et al. 2018

Preprint

View full text Add to dashboard Cite

Brain-computer interfaces (BCIs) have been largely developed to allow communication, control, and neurofeedback in human beings. Despite their great potential, BCIs perform inconsistently across individuals and the neural processes that enable humans to achieve good control remain poorly understood. To address this question, we performed simultaneous high-density electroencephalographic (EEG) and magnetoencephalographic (MEG) recordings in a motor imagery-based BCI training involving a group of healthy subjects.After reconstructing the signals at the cortical level, we showed that the reinforcement of motor-related activity during the BCI skill acquisition is paralleled by a progressive disconnection of associative areas which were not directly targeted during the experiments. Notably, these network connectivity changes reflected growing automaticity associated with BCI performance and predicted future learning rate. Altogether, our findings provide new insights into the large-scale cortical organizational mechanisms underlying BCI learning, which have implications for the improvement of this technology in a broad range of real-life applications.

show abstract

Section: Introductionmentioning

confidence: 99%

Spatiotemporal neural correlates of brain-computer interface learning

Corsi

Chávez

Schwartz

et al. 2018

Preprint

View full text Add to dashboard Cite

show abstract

“…Deep learning methods were originally developed in the computer vision field [48], and recently gained popularity in EEG analysis, in which they are used with the aim of improving classification performance over more traditional approaches, such as linear discriminant analysis, k-nearest neighbours or SVMs [20]. ConvNets are a type of feed-forward deep learning networks that are useful when data have a known topological structure [19,25].…”

Section: Methodsological Aspects Of Movement Predictionmentioning

confidence: 99%

“…Thus, pattern recognition systems used to decode and predict movements require careful engineering and domain expertise to transform raw EEG signals (usually by means of a feature extraction subsystem) into a suitable representation for the classification stage [19]. In this regard, several techniques have been proposed for feature extraction, e.g., common spatial patterns, independent component analysis, and joint timefrequency analysis, and also for classification, e.g., nearest neighbour classifier, linear discriminant analysis, support vector machines (SVMs), and ensemble strategies, among others [20]. An alternative is to use representation learning methods that automatically perform a feature extraction and classification through optimisation algorithms.…”

Section: Introductionmentioning

confidence: 99%

Prediction of Hand Movement Speed and Force from Single-trial EEG with Convolutional Neural Networks

Gatti

Atum

Schiaffino

et al. 2018

Preprint

View full text Add to dashboard Cite

Building accurate movement decoding models from brain signals is crucial for many biomedical applications. Decoding specific movement features, such as speed and force, may provide additional useful information at the expense of increasing the complexity of the decoding problem. Recent attempts to predict movement speed and force from the electroencephalogram (EEG) achieved classification accuracy levels not better than chance, stressing the demand for more accurate prediction strategies. Thus, the aim of this study was to improve the prediction accuracy of hand movement speed and force from single-trial EEG signals recorded from healthy volunteers. A strategy based on convolutional neural networks (ConvNets) was tested, since it has previously shown good performance in the classification of EEG signals. ConvNets achieved an overall accuracy of 84% in the classification of two different levels of speed and force (4class classification) from single-trial EEG. These results represent a substantial improvement over previously reported results, suggesting that hand movement speed and force can be accurately predicted from single-trial EEG.

show abstract

“…Briefly put, an RL agent takes states and rewards as inputs and yields an output action at a given time-step. States correspond to the binned spike firing rates sensed from micro-electrode arrays whereas rewards are scalar values obtained as a result of action taken at the previous **This work was supported through grant RG87/ 16 time-step. The notion of learning involves maximizing the score of total reward.…”

Section: Introductionmentioning

confidence: 99%

Towards Autonomous Intra-cortical Brain Machine Interfaces: Applying Bandit Algorithms for Online Reinforcement Learning

Shaikh

Sibindi

et al. 2020

Preprint

View full text Add to dashboard Cite

This paper presents application of Banditron -an online reinforcement learning algorithm (RL) in a discrete state intra-cortical Brain Machine Interface (iBMI) setting. We have analyzed two datasets from non-human primates (NHPs) -NHP A and NHP B each performing a 4-option discrete control task over a total of 8 days. Results show average improvements of ≈ 15%, 6% in NHP A and 15%, 21% in NHP B over state of the art algorithms -Hebbian Reinforcement Learning (HRL) and Attention Gated Reinforcement Learning (AGREL) respectively. Apart from yielding a superior decoding performance, Banditron is also the most computationally friendly as it requires two orders of magnitude less multiply-and-accumulate operations than HRL and AGREL. Furthermore, Banditron provides average improvements of at least 40%, 15% in NHPs A, B respectively compared to popularly employed supervised methods -LDA, SVM across test days. These results pave the way towards an alternate paradigm of temporally robust hardware friendly reinforcement learning based iBMIs.

show abstract

A review of classification algorithms for EEG-based brain–computer interfaces: a 10 year update

Abstract: This paper provides a comprehensive overview of the modern classification algorithms used in EEG-based BCIs, presents the principles of these methods and guidelines on when and how to use them. It also identifies a number of challenges to further advance EEG classification in BCI.

Cited by 1,439 publications

References 236 publications

Spatiotemporal neural correlates of brain-computer interface learning

Spatiotemporal neural correlates of brain-computer interface learning

Prediction of Hand Movement Speed and Force from Single-trial EEG with Convolutional Neural Networks

Towards Autonomous Intra-cortical Brain Machine Interfaces: Applying Bandit Algorithms for Online Reinforcement Learning

Contact Info

Product

Resources

About