Space & aviationResearch findings
A Method for Generating Synthetic IMU Data Using Bio-PM Representation Space
Original published 2026.11.20 03:56 KSTAdded to Blackdeck 2026.11.23 01:15 KST
What has changed?
A method is proposed to generate synthetic IMU data by leveraging token sequences in the Bio-PM representation space.
Key points
- Collecting IMU data for human activity recognition is costly and limits generalization across individuals.
- Synthetic data is generated in Bio-PM's representation space using token sequences, enabling effective cross-subject augmentation.
- Combining synthetic tokens with gravity-aware features improves leave-one-subject-out F1 score to 0.809 on WISDM benchmark.
Connect it to your work
Generated tokens exhibit realistic distributions and improve cross-subject generalization in activity recognition.
What remains unproven
This research note is based on an abstract; check the original for test conditions and application scope.
Originals used for this note
Links open the original publisher in a new tab.