Transfer Learning in Trajectory Decoding: Sensor or Source Space?
In this study, across-participant and across-session transfer learning was investigated to minimize the calibration time of the brain–computer interface (BCI) system in the context of continuous hand trajectory decoding. We reanalyzed data from a study with 10 able-bodied participants across three s...
| Main Authors: | , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
MDPI AG
2023-03-01
|
| Series: | Sensors |
| Subjects: | |
| Online Access: | https://www.mdpi.com/1424-8220/23/7/3593 |
| _version_ | 1797607083652153344 |
|---|---|
| author | Nitikorn Srisrisawang Gernot R. Müller-Putz |
| author_facet | Nitikorn Srisrisawang Gernot R. Müller-Putz |
| author_sort | Nitikorn Srisrisawang |
| collection | DOAJ |
| description | In this study, across-participant and across-session transfer learning was investigated to minimize the calibration time of the brain–computer interface (BCI) system in the context of continuous hand trajectory decoding. We reanalyzed data from a study with 10 able-bodied participants across three sessions. A leave-one-participant-out (LOPO) model was utilized as a starting model. Recursive exponentially weighted partial least squares regression (REW-PLS) was employed to overcome the memory limitation due to the large pool of training data. We considered four scenarios: generalized with no update (Gen), generalized with cumulative update (GenC), and individual models with cumulative (IndC) and non-cumulative (Ind) updates, with each one trained with sensor-space features or source-space features. The decoding performance in generalized models (Gen and GenC) was lower than the chance level. In individual models, the cumulative update (IndC) showed no significant improvement over the non-cumulative model (Ind). The performance showed the decoder’s incapability to generalize across participants and sessions in this task. The results suggested that the best correlation could be achieved with the sensor-space individual model, despite additional anatomical information in the source-space features. The decoding pattern showed a more localized pattern around the precuneus over three sessions in Ind models. |
| first_indexed | 2024-03-11T05:25:18Z |
| format | Article |
| id | doaj.art-e554386e7c7345149d4dc78fb3576765 |
| institution | Directory Open Access Journal |
| issn | 1424-8220 |
| language | English |
| last_indexed | 2024-03-11T05:25:18Z |
| publishDate | 2023-03-01 |
| publisher | MDPI AG |
| record_format | Article |
| series | Sensors |
| spelling | doaj.art-e554386e7c7345149d4dc78fb35767652023-11-17T17:34:52ZengMDPI AGSensors1424-82202023-03-01237359310.3390/s23073593Transfer Learning in Trajectory Decoding: Sensor or Source Space?Nitikorn Srisrisawang0Gernot R. Müller-Putz1Institute of Neural Engineering, Graz University of Technology, Stremayrgasse 16/IV, 8010 Graz, AustriaInstitute of Neural Engineering, Graz University of Technology, Stremayrgasse 16/IV, 8010 Graz, AustriaIn this study, across-participant and across-session transfer learning was investigated to minimize the calibration time of the brain–computer interface (BCI) system in the context of continuous hand trajectory decoding. We reanalyzed data from a study with 10 able-bodied participants across three sessions. A leave-one-participant-out (LOPO) model was utilized as a starting model. Recursive exponentially weighted partial least squares regression (REW-PLS) was employed to overcome the memory limitation due to the large pool of training data. We considered four scenarios: generalized with no update (Gen), generalized with cumulative update (GenC), and individual models with cumulative (IndC) and non-cumulative (Ind) updates, with each one trained with sensor-space features or source-space features. The decoding performance in generalized models (Gen and GenC) was lower than the chance level. In individual models, the cumulative update (IndC) showed no significant improvement over the non-cumulative model (Ind). The performance showed the decoder’s incapability to generalize across participants and sessions in this task. The results suggested that the best correlation could be achieved with the sensor-space individual model, despite additional anatomical information in the source-space features. The decoding pattern showed a more localized pattern around the precuneus over three sessions in Ind models.https://www.mdpi.com/1424-8220/23/7/3593electroencephalographybrain–computer interfacesource localizationtrajectory decodingpartial least-squares regressionunscented Kalman filter |
| spellingShingle | Nitikorn Srisrisawang Gernot R. Müller-Putz Transfer Learning in Trajectory Decoding: Sensor or Source Space? Sensors electroencephalography brain–computer interface source localization trajectory decoding partial least-squares regression unscented Kalman filter |
| title | Transfer Learning in Trajectory Decoding: Sensor or Source Space? |
| title_full | Transfer Learning in Trajectory Decoding: Sensor or Source Space? |
| title_fullStr | Transfer Learning in Trajectory Decoding: Sensor or Source Space? |
| title_full_unstemmed | Transfer Learning in Trajectory Decoding: Sensor or Source Space? |
| title_short | Transfer Learning in Trajectory Decoding: Sensor or Source Space? |
| title_sort | transfer learning in trajectory decoding sensor or source space |
| topic | electroencephalography brain–computer interface source localization trajectory decoding partial least-squares regression unscented Kalman filter |
| url | https://www.mdpi.com/1424-8220/23/7/3593 |
| work_keys_str_mv | AT nitikornsrisrisawang transferlearningintrajectorydecodingsensororsourcespace AT gernotrmullerputz transferlearningintrajectorydecodingsensororsourcespace |