Towards adaptive classification for BCI

By Pradeep Shenoy, Matthias Krauledat, Benjamin Blankertz, Rajesh P. N. Rao, K. Müller. Journal of Neural Engineering, 2006

484 citations recorded by the source corpus.

Subfields: Cognitive Neuroscience, Signal Processing

View Paper: https://www.semanticscholar.org/paper/f6003b2812e523d73854da1e7833202b54729bef

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Cognitive Neuroscience

In Signal Processing

Citations in the corpus

Introduction to machine learning for brain imaging (2011, methodology): “Moreover, a new generation of source localization techniques is now in use (e.g. Haufe et al., 2008, 2009) and has been successfully applied in BCI research (e.g. Pun et al., 2006; Noirhomme et al., 2008; Grosse-Wentrup et al., 2007).”
Optimizing Spatial filters for Robust EEG Single-Trial Analysis (2008, background): “With respect to CSP-based BCIs, the result of [29], [47] was that a simple adaptation of the classifier bias can compensate nonstationarity astonishingly well.”
Motor imagery and action observation: modulation of sensorimotor brain rhythms during mental control of a brain-computer interface. (2009, background): “This result is of interest in the context of a recent BCI study addressing the issue of neurophysiologic changes related to offline and online sessions (Shenoy et al., 2006).”
Single-trial analysis and classification of ERP components - A tutorial (2011, unknown): “However, regardless of the particular techniques employed for feature extraction or classification, there is substantial variability in the classification accuracy both between subjects [26, 7, 30, 25, 17, 1] and within subjects during the course of an experiment [62].”
A review of classification algorithms for EEG-based brain–computer interfaces: a 10 year update (2018, background): “With supervised adaptation, the true class labels of the incoming EEG signals is known and the classifier is retrained on the available training data augmented with these new, labelled incoming data, or is updated based on this new data only [200, 202].”
Review of the BCI Competition IV (2012, methodology): “In the context of BCI, it is mostly observed during the transition from the initial calibration phase to the online use of a BCI (Shenoy et al., 2006; Sugiyama et al., 2007), but also within periods of online use, where no obvious change of the task or paradigm takes place.”
Combining Brain–Computer Interfaces and Assistive Technologies: State-of-the-Art and Challenges (2010, background): “[173] P.”
Time Domain Parameters as a feature for EEG-based Brain-Computer Interfaces (2009, unknown): “Further improvement can be expected by adapting the bias, in a supervised manner like in Shenoy et al. (2006) or in an unsupervised fashion”
A brain-actuated wheelchair: asynchronous and non-invasive Brain-computer interfaces for continuous control of robots. (2008, unknown): “Shenoy and co-workers (Shenoy et al., 2006) describe two sources of non-stationarity, namely differences between samples extracted from calibration measurements (training data set) and samples extracted during the online operation of the BCI system (test data set), and changes in the user's brain…”
Brain Computer Interfaces, a Review (2012, unknown): “[214] described two main sources of non-stationarity.”
Towards a Cure for BCI Illiteracy (2009, methodology): “2) was updated for observation x ð t Þ using a recursive-least-square algorithm, (see Vidaurre et al. 2006 for more information): where v ð t Þ ¼ R (cid:4) 1 ð t (cid:4) 1 Þ (cid:2) x ð t Þ .”
A High-Speed Brain Speller using steady-State Visual evoked potentials (2014, methodology): “If adaptive learning approaches 38 are applicable, the training procedure could be further simplified by adopting training data from previous recordings.”
The non-invasive Berlin Brain-Computer Interface: Fast acquisition of effective performance in untrained subjects (2007, methodology): “Our investigations show that this adjustment is needed to account for the different experimental and mental conditions of the more demanding feedback situation when compared to the calibration session (cf. Krauledat et al., 2006; Shenoy et al., 2006).”
Signal Processing Approaches to Minimize or Suppress Calibration Time in Oscillatory Activity-Based Brain–Computer Interfaces (2015, methodology): “…data is available to calibrate the BCI for a given user, another approach that can be used to still reach reasonable classification performance is to use an adaptive approach [46], [47], i.e., to refine the calibration of the BCI during use, as new EEG data (labeled or not) become available.”
Covariate Shift Adaptation by Importance Weighted Cross Validation (2007, background): “On the other hand, strong non-stationarity effects have been often observed in brain signals between training and test sessions (Vidaurre et al., 2004; Millán, 2004; Shenoy et al., 2006), which could be regarded as an example of the covariate shift.”