AI Paper Atlas

Research map / Probabilistic and scientific modeling / Molecules and proteins

Cardio-mmFlow: A Gaussian-Prior-Free Physics-Informed Flow Matching Framework for Electrocardiogram to mmWave Radar Synthesis

Ziyang LIU, Ruiqiang Xiao, Chang Huang, KIEREN YU, Siyuan HE, Kaishun Wu

ICML 2026 · Probabilistic and scientific modeling › Molecules and proteins · approach: Flow matching generative modeling

Read the paperCheck your idea against 137,217 papers

AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
initialize flow from ECG latent instead of Gaussian noise (ablation)Cardio-mmFlow · ablation validation setLatent-FID: 24.55 -> 14.98Improved
use MSD-FiLM instead of vanilla MLP-FiLM for metadata modulation (ablation)Cardio-mmFlow · ablation validation setLatent-FID: 18.273 -> 14.98Improved
use MSD-FiLM instead of AdaLN + crossAttn conditioning (ablation)Cardio-mmFlow · ablation validation setLatent-FID: 24.93 -> 14.98Improved
use MSD-FiLM instead of AdaLN + crossAttn conditioning (ablation)Cardio-mmFlow · ablation validation setMSE: 0.026 -> 0.027No clear effect
add spectral fidelity loss Lfreq (ablation)Cardio-mmFlow · ablation validation setPSD similarity: 0.819 -> 0.919Improved
add Wasserstein alignment loss LWD (ablation)Cardio-mmFlow · ablation validation setLatent-FID: 24.64 -> 14.98Improved
add classifier-free guidance (ablation)Cardio-mmFlow · ablation validation setLatent-FID: 18.11 -> 14.98Improved
add classifier-free guidance (ablation)Cardio-mmFlow · ablation validation setMSE: 0.029 -> 0.027No clear effect
train downstream AirECG on Cardio-mmFlow synthetic data instead of other generators' dataAirECG zero-shot mmWave-to-ECG · zero-shot, self-collected datasetPCC: 0.74 -> 0.79Improved
add 10-40% real target-user calibration data to zero-shot modelAirECG · few-shot calibrationPCC: +4.17% at 10%; +2.95% at 40%Improved
add 10-20% real calibration dataAirECG · few-shot calibrationRR error: +1.71% at 10%; +0.52% at 20%No clear effect
use Cardio-mmFlow over PPGflowECG for heart-rate fidelitysynthesis · radar synthesis testHR-AE (bpm): 6.00 -> 6.66Hurt

Closest papers by meaning

Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy