# Phenomenon-Graph JEPA: Label-Efficient Representation Learning for Contactless Cardiorespiratory Sensing

arXiv: https://arxiv.org/abs/2609.32928 · HTML: https://mlsanity.com/paper/2609.32928/

## Key facts

- No citing papers in the map yet (data through 14 Oct 2026).
- Builds on 9 of its 18 references that are in the map.
- First posted on arXiv 26 Sep 2026.

## Builds on (most cited first, 9 of 9)

- [Self-Supervised Learning from Images with a Joint-Embedding Predictive Architecture](https://mlsanity.com/paper/2301.08243/) (2023)
- [Time-Series Representation Learning via Temporal and Contextual Contrasting](https://mlsanity.com/paper/2106.14112/) (2021)
- [Data Augmentation techniques in time series domain: A survey and taxonomy](https://mlsanity.com/paper/2206.13508/) (2022)
- [Self-Supervised Pre-Training with Joint-Embedding Predictive Architecture Boosts ECG Classification Performance](https://mlsanity.com/paper/2410.13867/) (2024)
- [Joint Embeddings Go Temporal](https://mlsanity.com/paper/2509.25449/) (2025)
- [Giving Sensors a Voice: Multimodal JEPA for Semantic Time-Series Embeddings](https://mlsanity.com/paper/2605.31580/) (2026)
- [OMuSense-23: A Multimodal Dataset for Contactless Breathing Pattern Recognition and Biometric Analysis](https://mlsanity.com/paper/2407.06137/) (2024)
- [CardioState-JEPA: Delay-Aware Cross-Modal Learning of a Shared Cardiac Representation](https://mlsanity.com/paper/2608.12944/) (2026)
- [CF-JEPA: Mask-free forward prediction with asymmetric encoder utilization for time-series representation learning](https://mlsanity.com/paper/2606.07031/) (2026)

## Abstract (arXiv)

Millimeter-wave (mmWave) radar and RGB-D cameras can record cardiac and respiratory waveforms continuously and without contact, but labeled recordings remain scarce because every label requires a supervised acquisition session. Self-supervised pretraining can exploit the unlabeled signals, yet contrastive methods depend on signal transformations and negative pairs whose validity is uncertain for cardiorespiratory data, where time warping changes breathing rate and distant windows can share the same physiological state. We present Phenomenon-Graph JEPA, a joint-embedding predictive architecture that learns from four processed one-dimensional streams without negative pairs or synthetic augmentation in its base configuration. Each stream is encoded by a temporal convolutional branch and a band-limited spectral branch. During pretraining, the model predicts stopped target embeddings along typed edges, which connect streams assigned to the same physiological phenomenon, and forward in time within a state episode. We treat this physiological typing as a testable hypothesis and compare it with wrong-edge and all-pairs prediction graphs. In the OMuSense-23 dataset, pretraining improves label-efficiency area over matched supervised training by 3.91 percentage points (95% interval 2.08 to 5.80, Holm-adjusted p = 0.006), and by 3.74 points under a second configuration evaluated on the same test participants. However, the wrong-edge and all-pairs controls do not establish a benefit from physiological typing. Optional Takens-inspired delay coordinates improve a validation comparison with learned history, whereas two wrist-only WESAD protocols do not establish a pretraining advantage. The study therefore separates the measured benefit of predictive representations from the physiological prior used to organize their training.