Cross-modal latent alignment enables efficient compression of wearable sEMG and accelerometer signals
Yongfei Liu, Chengwen Liu, Ruili Chang, Zhiyou Zhang, Yinghao Zhang · 2026
Read Cross-modal latent alignment enables efficient compression of wearable sEMG and accelerometer signals by Yongfei Liu, Chengwen Liu, Ruili Chang, Zhiyou Zhang, Yinghao Zhang (2026) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/pmc-13473458.txt.
Description
Yongfei Liu and four colleagues proposed in 2026 a smarter way to compress signals from wearable sensors that track muscles and movement. Surface electromyography, which records muscle electrical activity through the skin, and accelerometers, which record limb motion, are naturally linked, since muscle activation drives movement, yet current compressors handle each signal alone. Their framework aligns the internal representations of the two signals during training. Matching frames and temporal patterns teaches the compressor their shared structure.
Tests on two Ninapro datasets showed gains across compression ratios from 20 to 90 percent. Correlation ran up to 28 percent higher than traditional methods, and signal quality ran 1 to 3 decibels above deep learning baselines. Gesture recognition stayed above 80 percent accurate at 40 percent compression, a level rivals needed 50 percent or more to match. Because the alignment module is dropped after training, the deployed system costs no extra computing, a point that matters for small, battery-powered wearables.
Catalog record
| CREATOR | Yongfei Liu, Chengwen Liu, Ruili Chang, Zhiyou Zhang, Yinghao Zhang |
|---|---|
| YEAR | 2026 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 58 KB |
| VERSION | 1.0 |
| LICENSE | CC BY 4.0 |
| RIGHTS | CC BY |
| SOURCE | Europe PMC |