eth-siplab/Learning-with-FrameProjections

Learning Without Augmenting — Preprocessed Time-Series Data

Preprocessed data used in our NeurIPS 2025 paper Learning Without Augmenting.
It includes nine datasets across five time-series tasks in ready-to-use format.

[Hugging Face Papers]

Code

[Project Page]

Datasets Included

File	Format	Description
Dalia_data.pkl	PKL	Large-scale benchmark for heartrate estimation using wearables
ECG_data.pkl	PKL	ECG (combined) for cardiovascular disease classification
HHAR.zip	ZIP	Human activity dataset with IMUs
IEEE_Big.mat	MAT	Heartrate estimation in a controlled environment
IEEE_Small.mat	MAT	A smaller version of the IEEE dataset
clemson.mat	MAT	IMU data collected from wearable devices for step counting
sleep_combined.pt	PKL	Sleep stage classification data

---