Calibration-Free BCI Based Control

Authors: Jonathan Grizou, Iñaki Iturrate, Luis Montesano, Pierre-Yves Oudeyer, Manuel Lopes

AAAI 2014 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental 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.
Researcher Affiliation Academia Jonathan Grizou Inria Bordeaux Sud-Ouest, France jonathan.grizou@inria.fr Inaki Iturrate CNBI, EPFL, Switzerland inaki.iturrate@epfl.ch Luis Montesano I3A, Univ. of Zaragoza, Spain montesano@unizar.es Pierre-Yves Oudeyer Inria Bordeaux Sud-Ouest, France pierre-yves.oudeyer@inria.fr Manuel Lopes Inria Bordeaux Sud-Ouest, France manuel.lopes@inria.fr
Pseudocode No The paper describes the algorithm formally through equations and prose but does not include any explicit pseudocode blocks or sections labeled "Algorithm".
Open Source Code No The paper does not contain any explicit statements about releasing source code for the described methodology or links to a code repository.
Open Datasets Yes We used a dataset from (Iturrate, Montesano, and Minguez 2013b), which covers ten subjects that performed two different control problems (denoted T1 and T2).
Dataset Splits No The paper mentions "cross-validation" as a possible solution for estimating classification rates but does not provide specific train/validation/test dataset splits (percentages or counts) used for their experiments.
Hardware Specification No The paper mentions "EEG signals were recorded with a g Tec system (2 g USBamp amplifiers)" for data acquisition, but it does not provide specific hardware details (like GPU/CPU models, processor types, or memory amounts) used for running the computational experiments.
Software Dependencies No The paper does not provide specific ancillary software details, such as library or solver names with version numbers, needed to replicate the experiment.
Experiment Setup Yes The user goal is to teach the agent to reach one, yet unknown to the agent, of the 25 discrete states..., EEG signals were recorded with a g Tec system..., sampling frequency of 256 Hz, common-average-reference (CAR) filtered and band-pass filtered at [0.5, 10] Hz., features were extracted from two frontocentral channels (FCz and Cz) within a time window of [200, 700] ms... and downsampled to 32 Hz. This leaded to a vector of 34 features., modeling the assessment error rate of the user, which was set to 0.1 for our experiments., λ the regularization term which was set to 0.5 for our experiments., We used β = 0.99., In practice, we limited D t i to the last 250 elements.