From full calibration to zero training for a code-modulated visual evoked potentials brain computer interface

Thielen, Jordy; Marsman, Pieter; Farquhar, Jason; Desain, Peter

doi:10.1088/1741-2552/abecef

Cited by 24 publications

(50 citation statements)

References 38 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…An unsupervised online adaptive method could ensure the development of a high-performance online BCI system. Therefore, in the future, the ALPHA’s feature extraction methods could be applied to the code-modulated-VEP (c-VEP)-based BCI system since both c-VEP and the SSVEP have common feature information ( Thielen et al, 2021 ).…”

Section: Discussionmentioning

confidence: 99%

Facilitating Applications of SSVEP-Based BCIs by Within-Subject Information Transfer

Liu

Gao

et al. 2022

Front. Neurosci.

View full text Add to dashboard Cite

The steady-state visual evoked potential based brain–computer interface (SSVEP–BCI) can provide high-speed alternative and augmentative communication in real-world applications. For individuals using a long-term BCI, within-subject (i.e., cross-day and cross-electrode) transfer learning could improve the BCI performance and reduce the calibration burden. To validate the within-subject transfer learning scheme, this study designs a 40-target SSVEP–BCI. Sixteen subjects are recruited, each of whom has performed experiments on three different days and has undergone the experiments of the SSVEP–BCIs based on the dry and wet electrodes. Several transfer directions, including the cross-day directions in parallel with the cross-electrode directions, are analyzed, and it is found that the transfer learning-based approach can maintain stable performance by zero training. Compared with the fully calibrated approaches, the transfer learning-based approach can achieve significantly better or comparable performance in different transfer directions. This result verifies that the transfer learning-based scheme is well suited for implementing a high-speed zero-training SSVEP–BCI, especially the dry electrode-based SSVEP–BCI system. A validation experiment of the cross-day wet-to-dry transfer, involving nine subjects, has shown that the average accuracy is 85.97 ± 5.60% for the wet-to-dry transfer and 77.69 ± 6.42% for the fully calibrated method with dry electrodes. By leveraging the electroencephalography data acquired on different days by different electrodes via transfer learning, this study lays the foundation for facilitating the long-term usage of the SSVEP–BCI and advancing the frontier of the dry electrode-based SSVEP–BCI in real-world applications.

show abstract

Section: Discussionmentioning

confidence: 99%

Facilitating Applications of SSVEP-Based BCIs by Within-Subject Information Transfer

Liu

Gao

et al. 2022

Front. Neurosci.

View full text Add to dashboard Cite

show abstract

“…Compared to the system developed by Thielen et al [ 26 ], which only requires an initial warm-up period of 12 s, the here-presented approach requires a minimum of one new training block, which corresponds to 98.2 s (including the inter-trial time of 1.0 s). On the other hand, some minimal training can be beneficial (easier adaptation) for different stimulation hardware (higher refresh rates which corresponds to higher cVEP carrier frequency), where the number of gold codes can be limited; instead, longer m-sequences can be implemented: one example is a 124-bit quintary m-sequence [ 33 ]).…”

Section: Discussionmentioning

confidence: 99%

“…Our findings show that at least two training blocks from two sessions form a sufficient starting point for further improvements in a modern cVEP-based BCI system, which in future research can further be improved with e.g., an online adaptation process. While the novel developed/emerging systems can work without a user-specific training (e.g., Spüler et al [ 9 ] or Thielen et al [ 26 ]), most of the currently used and popular cVEP-based BCI systems require some training phase that could benefit from previously collected training data, also from other subjects. This cross-subject feature extraction requires further research.…”

Section: Discussionmentioning

confidence: 99%

“…Recently developed novel cVEP-based BCI systems focus on skipping the mostly required training session [ 25 , 26 ]. In 2018, Nagel and Spüler developed a model that predicts the arbitrary visual stimuli from EEG and the triggered brain response [ 25 ].…”

Section: Introductionmentioning

confidence: 99%

“…In an online experiment, nine participants achieved an average accuracy and ITR of 98.0% and 108.1 bits/min, respectively, for an optimised EEG2Code model using 32 EEG electrodes. In the recent paper, Thielen et al developed a calibration-free method (0-training approach), where in an online copy-spelling task, nine participants achieved an average accuracy and ITR of 99.7% and 67.9 bits/min, respectively, using eight EEG electrodes and gold codes-based stimuli [ 26 ].…”

Section: Introductionmentioning

confidence: 99%

See 2 more Smart Citations

cVEP Training Data Validation—Towards Optimal Training Set Composition from Multi-Day Data

Stawicki

Volosyak

2022

Brain Sciences

View full text Add to dashboard Cite

This paper investigates the effects of the repetitive block-wise training process on the classification accuracy for a code-modulated visual evoked potentials (cVEP)-based brain–computer interface (BCI). The cVEP-based BCIs are popular thanks to their autocorrelation feature. The cVEP-based stimuli are generated by a specific code pattern, usually the m-sequence, which is phase-shifted between the individual targets. Typically, the cVEP classification requires a subject-specific template (individually created from the user’s own pre-recorded EEG responses to the same stimulus target), which is compared to the incoming electroencephalography (EEG) data, using the correlation algorithms. The amount of the collected user training data determines the accuracy of the system. In this offline study, previously recorded EEG data collected during an online experiment with 10 participants from multiple sessions were used. A template matching target identification, with similar models as the task-related component analysis (TRCA), was used for target classification. The spatial filter was generated by the canonical correlation analysis (CCA). When comparing the training models from one session with the same session’s data (intra-session) and the model from one session with the data from the other session (inter-session), the accuracies were (94.84%, 94.53%) and (76.67%, 77.34%) for intra-sessions and inter-sessions, respectively. In order to investigate the most reliable configuration for accurate classification, the training data blocks from different sessions (days) were compared interchangeably. In the best training set composition, the participants achieved an average accuracy of 82.66% for models based only on two training blocks from two different sessions. Similarly, at least five blocks were necessary for the average accuracy to exceed 90%. The presented method can further improve cVEP-based BCI performance by reusing previously recorded training data.

show abstract

Effects of Stimulus Sequences on Brain-Computer Interfaces Using Code-Modulated Visual Evoked Potentials: An Offline Simulation

Thielen

2023

Lecture Notes in Computer Science

View full text Add to dashboard Cite

Brain-computer interfaces (BCIs) translate brain activity into computer commands opening a novel non-muscular channel for communication and control. Before BCIs can be used effectively, they require extensive calibration of their machine learning algorithms using labeled training data. This tedious calibration process is not only timeconsuming and costly but also restricts the exploration and optimization of stimulus parameters that could greatly enhance BCI performance. To overcome the challenge of acquiring large training datasets, a simulation framework was developed to eliminate the need for recording calibration data. Unlike previous studies, this simulation framework incorporates a biologically plausible forward model of the code-modulated visual evoked potential (c-VEP). By utilizing synthetic data generated by this improved simulation framework, an offline study was conducted to systematically compare five different stimulus conditions: almost perfect autocorrelation (APA) sequence, de Bruijn sequence, Golay sequence, Gold code, and m-sequence. The results of the study revealed that the Golay sequence achieved the highest grand average performance, followed by the APA sequence, m-sequence, Gold code, and finally the de Bruijn sequence. Furthermore, when the stimulus sequence was optimized for individual participants, typically the Golay and APA sequences exhibited the highest classification accuracy, leading to a significant improvement in overall accuracy as compared to using the Golay sequence for all participants. This research represents an important initial step towards optimizing stimulus parameters for BCIs using simulated data. This approach has the potential to accelerate the development and optimization of BCIs, resulting in more effective applications that can be tailored to individual users.

show abstract

From full calibration to zero training for a code-modulated visual evoked potentials brain computer interface

Cited by 24 publications

References 38 publications

Facilitating Applications of SSVEP-Based BCIs by Within-Subject Information Transfer

Facilitating Applications of SSVEP-Based BCIs by Within-Subject Information Transfer

cVEP Training Data Validation—Towards Optimal Training Set Composition from Multi-Day Data

Effects of Stimulus Sequences on Brain-Computer Interfaces Using Code-Modulated Visual Evoked Potentials: An Offline Simulation

Contact Info

Product

Resources

About