Research map / 3D vision / 3D generation
XSleepNet: Multi-View Sequential Model for Automatic Sleep Staging
Huy Phan, Oliver Y. Chén, Minh C. Tran, Philipp Koch, Alfred Mertins, Maarten De Vos
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 changed | Where | Measured | Effect |
|---|---|---|---|
| add adaptive gradient blending to naive concatenation fusion (XSleepNet2) (ablation) | Naive Fusion · SleepEDF-20 (+-30 min, EEG) | accuracy (%): 85.0 -> 86.3 | Improved |
| add adaptive gradient blending to naive concatenation fusion (XSleepNet2) (ablation) | Naive Fusion · SleepEDF-20 (in-bed, EEG) | accuracy (%): 80.2 -> 83.9 | Improved |
| add adaptive gradient blending to naive concatenation fusion (XSleepNet2) (ablation) | Naive Fusion · MASS (EEG, EOG, EMG) | accuracy (%): 86.8 -> 87.6 | Improved |
| add adaptive gradient blending (XSleepNet1) to naive fusion on average over databases | Naive Fusion · all five databases | accuracy (%): +0.9% absolute | Improved |
| add adaptive gradient blending (XSleepNet2) to naive fusion on average over databases | Naive Fusion · all five databases | accuracy (%): +1.2% absolute | Improved |
| add raw-signal stream to time-frequency stream via XSleepNet2 vs ARNN+RNN average | ARNN+RNN · all five databases | accuracy (%): +0.9% absolute | Improved |
| add time-frequency stream to raw stream via XSleepNet2 vs FCNN+RNN average | FCNN+RNN · all five databases | accuracy (%): +1.4% absolute | Improved |
| combine views with naive concatenation instead of using best single view | Naive Fusion · SleepEDF-20 (in-bed, EEG) | accuracy (%): 82.2 -> 80.2 | Hurt |
| use second-order approximation (XSleepNet2) instead of first-order (XSleepNet1) for loss weights (ablation) | XSleepNet1 · SleepEDF-20 (in-bed, EEG) | accuracy (%): 82.2 -> 83.9 | Improved |
| use second-order approximation (XSleepNet2) instead of first-order (XSleepNet1) with EEG+EOG (ablation) | XSleepNet1 · SleepEDF-20 (+-30 min, EEG+EOG) | accuracy (%): 85.2 -> 86.4 | Improved |
| use XSleepNet1 instead of XSleepNet2 with EEG+EOG+EMG | XSleepNet2 · MASS (EEG, EOG, EMG) | accuracy (%): 87.6 -> 87.5 | No clear effect |
| pretrain on MASS and fine-tune on SleepEDF-20 (FT-XSleepNet) vs no transfer | XSleepNet · SleepEDF-20 (in-bed, EEG) | accuracy (%): 83.9 -> 85.7 | Improved |
Closest papers by meaning
- Generalizable Sleep Staging via Multi-Level Domain Alignment (AAAI 2024 · similarity 0.90)
- Personalized Sleep Staging Leveraging Source-free Unsupervised Domain Adaptation (AAAI 2025 · similarity 0.88)
- SleepFM: Multi-modal Representation Learning for Sleep Across Brain Activity, ECG and Respiratory Signals (ICML 2024 · similarity 0.88)
- Robust Sleep Staging over Incomplete Multimodal Physiological Signals via Contrastive Imagination (NeurIPS 2024 · similarity 0.87)
- Subject-Relative Micro-Motion and Sleep Dynamics for Near-Infrared Video Sleep Staging (NeurIPS 2026 · similarity 0.86)
- SleepVST: Sleep Staging from Near-Infrared Video Signals using Pre-Trained Transformers (CVPR 2024 · similarity 0.83)
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.
