Computational Model Studio · Model release
NormWear
Encodes multichannel physiological wearable sensor signals into reusable signal representations.
Developer: Mobile Sensing and Ubiquitous Computing Laboratory · Last verified:
What it is designed to do
Encode multivariate wearable physiological sensor waveforms into per-channel and per-patch embeddings for downstream research tasks.
- Primary contribution
- representation
- Input
- sensor, time series
- Output
- number vector (per-channel signal embeddings, per-patch signal embeddings)
Documented uses: Extract representations from multichannel PPG, ECG, EEG, GSR and IMU waveforms for research evaluation.
Access and deployment
- NormWear model.safetensors weights — available. Public publisher model repository provides model.safetensors and custom model-loading code. Artifact documentation Local hardware requirements have not been verified. Apache License 2.0.
Technical profile and review notes
- Construction
- learned
- Representation
- neural
- Normalized architecture tags
- No normalized architecture tag assigned
- Lifecycle
- Unknown from reviewed sources
Inclusion rationale: Adds a specific physiological sensor-stream input and embedding output that the prior working collection did not represent.
Limitations
- The card demonstrates embedding extraction; clinical diagnosis or individual health decisions are not established by this record.
- Inputs require compatible channels, sampling, and preprocessing; transfer performance on a new sensor setup is not independently established here.
- The publisher example loads custom repository code with trust_remote_code=True; the code and dependencies require separate review.
- No laptop, mobile-device, or throughput profile has been verified for this checkpoint.
Sources and related Studio records
Sources
Connections
- In areaComputational Models
- Supports functionEncode sensor signals
- ArtifactNormWear model.safetensors weights