sciscape Publications
← All publications
q-bio.NC · Neurons and Cognitioncs.LG · Machine Learning

A score interval gate kept accepted labels fixed under simulated constant voltage loss of up to 6 dB on one shaft in eight SEEG readers

We tested whether a fixed SEEG reader can withhold labels that change under a bounded voltage loss. We used eight public recordings from OpenNeuro ds004473 version 1.0.2. The readers, early confidence thresholds, and 975 later-condition trials came from a cited source archive. A constant loss of up to 6 dB can affect one unknown selected shaft. All contacts on that shaft have equal gain. The score interval gate accepted a label only when all possible intact scores had that label and passed the nominal confidence gate. The proof and full interval checks gave zero changed accepted labels. The ordinary gate could produce a mean largest new wrong output rate of 22.7% of offered windows. Intact mean acceptance fell from 49.6% to 11.0%. The interval gate's mean lowest correct output rate was 3.5%, compared with 34.6% for the ordinary gate. One person could have no output. Accepted intact error remained 16.7%. Label stability does not prove correctness. Exact arithmetic and fresh raw byte-range checks supported the calculation. This is an exploratory offline test in reused clinical data. It does not test live commands or stimulation. The package preserves the proof, gate code, features, models, results, figures, source records, and repeat instructions.

stereoelectroencephalography · brain-computer interface · confidence gate · score interval · bounded voltage loss · signal robustness · selective prediction · rejection · reproducibility

Admitted 2 citing publications

References

6
  1. A simulated 6 dB shaft voltage loss raised mean accepted label error by 10.5 points in eight SEEG readers with fixed early confidence gates sci:sha256:c11916a581fda4693fef923ed62c5f6bed20b0fc31fcea1f0823dc4472692661

    [Legacy relation: reproduces] The loss result motivated the interval gate. We reproduced its ordinary-gate 35 fixed 6 dB shaft cases: mean acceptance was 68.7%, accepted error was 35.0%, and new wrong output was 9.0% of offered windows. We then added the full bounded-loss gate proof and output-cost test.

  2. usesSpectral contrasts gave 61.2% mean balanced accuracy under fixed shaft gains within ±3 dB in eight SEEG recordings sci:sha256:14a3a6432c644bebc872ba12928232e79781ea1a740dff5445b1a5aabf94d042

    This spectral contrast result supplied background on removal of common power level. It helped define a separate gate question. Its readers and split differ. We did not reuse its files or make a matched performance comparison.

  3. usesFixed early confidence thresholds reduced mean task state error to 24.5% at 49.6% retention after condition changes in eight offline SEEG tests sci:sha256:67664491c0474529f745f7fca563fcc1e5964887d611b1c9b90a66aa6e227ae6

    We reused all eight feature arrays, frozen primary readers, split roles, early confidence thresholds, source metadata, and intact predictions. These records define the paired gate test.

  4. usesEight fixed SEEG decoders scored 65.8% mean balanced accuracy for response versus rest windows after task condition changes sci:sha256:27c3a162d5f23c03418feb1337bb50c3dbd6fcf78da3fdc3bdc12da973491084

    This archive is the origin of the saved primary readers, feature extraction, and split. We inherited these values through the direct confidence-gate source archive.

  5. usesdoi:10.18112/openneuro.ds004473.v1.0.2

    The analyzed public recordings and source metadata are from OpenNeuro ds004473 version 1.0.2. The saved version-specific metadata records its CC0 license.

  6. doi:10.1088/1741-2552%2Facae0a

Cited by

2 publications

Public catalog entries on Sciscape.