Search papers, labs, and topics across Lattice.
BioFormer addresses cross-subject generalization in biomedical time-series by explicitly modeling and suppressing subject-specific spectral variability. The core innovation is a Frequency-Band Alignment Module (FBAM) that adaptively adjusts the amplitude and phase of frequency components based on spectral distribution to align spectral structure across subjects. Experiments on six datasets show BioFormer outperforms 12 baselines, achieving a 6% absolute F1-score improvement.
Subject-specific variability in biomedical time-series can be mitigated by explicitly aligning spectral structure, leading to a 6% F1-score improvement over existing methods.
Cross-subject generalization in biomedical time-series refers to training on data from some subjects and testing on unseen subjects.The key challenge is to suppress subject specific variability in BTS representations.Most existing methods implicitly suppress the variability through model building or subject adversarial learning, but rarely model it explicitly.We introduce spectral drift as a new perspective to characterize subject specific variability.Specifically, BTS signals under the same label often share consistent oscillatory structure, yet exhibit subject-dependent magnitude or phase shifts in specific frequency components, which we interpret as subject-specific variability. Building on this insight, we propose BioFormer.At its core is a Frequency-Band Alignment Module(FBAM) that generates band-wise modulation factors from the spectral distribution and adaptively adjusts amplitude and phase to align spectral structure, thereby mitigating variability.We further pair FBAM with Sample Conditional Layer Normalization, which infers normalization parameters from intrinsic signal statistics rather than subject identity, stabilizing cross-subject representations.Extensive experiments on six datasets demonstrate that BioFormer outperforms 12 baselines, yielding absolute F1-score improvements of 6%.
