Search papers, labs, and topics across Lattice.
Affiliation:
2
0
1
0
Most current PPG foundation models fail to generalize, with over 80% not outperforming simple PRV features or generic time-series models.
Dense attention drowns low-SNR, multi-channel data in spurious cross-token noise; factorizing attention into orthogonal spatial and temporal paths with adaptive gating systematically outperforms isotropic Transformers on neural and audio signals.