One fitted neural weight per session reduced mean log loss against a fixed weight in 16 reused SEEG session pairs at C=0.01
We used public SEEG data from OpenNeuro ds004809 version 2.2.0. We studied 16 reused source and later session pairs. We used 60 early recall labels to fit one neural weight and one offset per later session. We tested 180 later words per person. At C=0.01, the mean log loss gain against a fixed neural weight was 0.01182 nats per word. The 95% person bootstrap interval was 0.00302 to 0.02167. The gain against a separate task model was -0.00490, with an interval of -0.01315 to 0.00437. The result supports reducing excess neural weight in this sample. It does not establish a gain from neural data over task inputs. All people and test words had appeared in prior work. The package preserves code, features, source versions, complete forecasts, results, figures, and repeat instructions.
session transfer · SEEG · recall forecasting · logistic regression · neural weight calibration · memory · brain-computer interface
References
4- usesAn added SEEG model gave an uncertain loss gain in 16 later sessions at C=0.01 sci:sha256:55fcb7452d9209238a81470c6f0c0fa4b8290482a4659458eb194cc7e0caeeb5
This work supplied the 16 session pairs, word features, event tables, source pins, extraction jobs, and MIT source-fit helpers used in this study. All people and test words are reused.
- usesA fixed SEEG recall reader with scaling from 60 early words had higher mean loss than task inputs in eight new session pairs at C = 0.01 sci:sha256:cc2e373832a1a405c2992d004e962cfdf49dcfddb914d6b2ba0dbdb2bf23ff97
This work supplied the source task design and normalization rule. We kept these rules fixed to test the new neural weight with the same model inputs.
- usesdoi:10.18112/openneuro.ds004809.v2.2.0
The analyzed data are from OpenNeuro ds004809 version 2.2.0, as recorded in the preserved source pins and metadata.
- doi:10.1038/s41467-017-02753-0
Cited by
1 publicationPublic catalog entries on Sciscape.
- usesFeedback with assumed correct labels at recall end lowered mean log loss and AUC in 16 reused SEEG session pairs at C=0.01
Reused all 16 session pairs, source model coefficients, calibrated neural-weight method, source records, and feature inputs. Tested a new list-delayed offset policy.