Partially supervised P300 speller adaptation for eventual stimulus timing optimization: target confidence is superior to error-related potential score as an uncertain label
Bibliographic record
Abstract
OBJECTIVE: Error-related potentials (ErrPs) have the potential to guide classifier adaptation in BCI spellers, for addressing non-stationary performance as well as for online optimization of system parameters, by providing imperfect or partial labels. However, the usefulness of ErrP-based labels for BCI adaptation has not been established in comparison to other partially supervised methods. Our objective is to make this comparison by retraining a two-step P300 speller on a subset of confident online trials using naïve labels taken from speller output, where confidence is determined either by (i) ErrP scores, (ii) posterior target scores derived from the P300 potential, or (iii) a hybrid of these scores. We further wish to evaluate the ability of partially supervised adaptation and retraining methods to adjust to a new stimulus-onset asynchrony (SOA), a necessary step towards online SOA optimization. APPROACH: Eleven consenting able-bodied adults attended three online spelling sessions on separate days with feedback in which SOAs were set at 160 ms (sessions 1 and 2) and 80 ms (session 3). A post hoc offline analysis and a simulated online analysis were performed on sessions two and three to compare multiple adaptation methods. Area under the curve (AUC) and symbols spelled per minute (SPM) were the primary outcome measures. MAIN RESULTS: Retraining using supervised labels confirmed improvements of 0.9 percentage points (session 2, p < 0.01) and 1.9 percentage points (session 3, p < 0.05) in AUC using same-day training data over using data from a previous day, which supports classifier adaptation in general. SIGNIFICANCE: Using posterior target score alone as a confidence measure resulted in the highest SPM of the partially supervised methods, indicating that ErrPs are not necessary to boost the performance of partially supervised adaptive classification. Partial supervision significantly improved SPM at a novel SOA, showing promise for eventual online SOA optimization.
Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.
How this classification was reachedexpand
Full frame distilled prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. Learned from the 10,348 direct Codex labels and 10,348 direct Gemma labels. Candidate is the union of thresholded teacher heads; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels or direct frontier model labels.
Codex and Gemma teacher scores by category
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.000 | 0.000 |
| Meta-epidemiology (narrow) | 0.000 | 0.000 |
| Meta-epidemiology (broad) | 0.000 | 0.000 |
| Bibliometrics | 0.000 | 0.000 |
| Science and technology studies | 0.000 | 0.000 |
| Scholarly communication | 0.000 | 0.001 |
| Open science | 0.000 | 0.000 |
| Research integrity | 0.000 | 0.000 |
| Insufficient payload (model declined to judge) | 0.000 | 0.000 |
Machine scores (provisional)
The two teacher heads of the student model, read on this work. A score orders the frame for review; it never asserts a category, and the validation status ships verbatim with every row.
Baseline scores from an immature model (maturity gate not passed, 7 training rounds). Scores rank; they never assert a category.
score_only:v0-immature-baseline · verbatim from the scoring run: score_only means the number may rank works, and no category label ships from itClassification
machine, unvalidatedMachine predicted; a candidate call from one teacher head, not a consensus.
How this classification was reached, model by model and score by score, is at the end of the page under "How this classification was reached".