Calibration-Free BCI Based Control

Abstract

Recent works have explored the use of brain signals to directly control virtual and robotic agents in sequential tasks. So far in such brain-computer interfaces (BCI), an explicit calibration phase was required to build a decoder that translates raw electroencephalography (EEG) signals from the brain of each user into meaningful instructions. This paper proposes a method that removes the calibration phase, and allows a user to control an agent to solve a sequential task. The proposed method assumes a distribution of possible tasks, and infers the interpretation of EEG signals and the task by selecting the hypothesis which best explains the history of interaction. We introduce a measure of uncertainty on the task and on the EEG signal interpretation to act as an exploratory bonus for a planning strategy. This speeds up learning by guiding the system to regions that better disambiguate among task hypotheses. We report experiments where four users use BCI to control an agent on a virtual world to reach a target without any previous calibration process.

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

“…This method exploited the specific structure of the P300-speller, and notably the frequency of samples from each class at each time, to estimate the probability of the most likely class label. In a related work, [78] proposed a generic method to adaptively estimate the parameters of the classifier without knowing the true class labels by exploiting any structure that the application may have. Semi-supervised adaptation was also used offline for multi-class motor imagery with a Kernel discriminant analysis (KDA) classifier in [171].…”

“…Nonetheless unsupervised adaptation has been shown to be superior to static classifiers in multiple studies [24,130,132,149,219]. It can also be used to shorten or even remove the need for calibration [78,81,105,122,151]. There is a need for more robust unsupervised adaptation methods, as the majority of actual BCI applications do not provide labels, and thus can only rely on unsupervised methods.…”

Advancing passive BCIs: a feasibility study of two temporal derivative features and effect size-based feature selection in continuous online EEG-based machine error detection

“…Christensen et al ( 2012 ) applied three popular pattern classification techniques to EEG data from participants executing a complex multi-task over a span of 5 days in a month, observing a significant decrease in classification accuracy across different days. Although adaptive strategies (Shenoy et al, 2006 ) and a few calibration-free online classification architectures (Grizou et al, 2014 ; Wimpff et al, 2023 ) have been proposed for offline to online session transfer, it is crucial to investigate more pattern recognition approaches in a calibration-free manner over longer periods, such as days or weeks. The second challenge is the presence of artifacts in EEG signals during the continuous online classification.…”

A Calibration-free Approach to Implementing P300-based Brain–computer Interface

“…In 2014, Kindermans et al [21] attempted to bypass the calibration recording by utilizing an unsupervised trained classifier, which was initialized randomly and then updated during usage. In the same year, Grizou et al [22] proposed a method without the need for calibration, which could contin-uously update the inference about the interpretation of EEG signals of BCI users as new data comes in. In 2015, Bauer et al [4] applied a Bayesian model of neurofeedback and reinforcement learning to study the impact of threshold adaptation of classifiers on optimizing restorative BCIs.…”