Research map / Probabilistic and scientific modeling / Molecules and proteins
NeuralFLoC: Neural Flow-Based Joint Registration and Clustering of Functional Data
Xinyang Xiong, Siyuan Jiang, Pengcheng Zeng
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 |
|---|---|---|---|
| Use Neural ODE warper instead of discrete SrvfRegNet in joint training (ablation) | NeuralFLoC · Wave (d=1) | ACC: 0.552 -> 0.993 | Improved |
| Use Neural ODE warper instead of SrvfRegNet (ablation) | NeuralFLoC · Shapes | ACC: 0.908 -> 0.937 | Improved |
| Use Neural ODE warper instead of SrvfRegNet (ablation) | NeuralFLoC · Symbols (2 classes) | ATV: 2.9 vs 3.1 | No clear effect |
| Add registration module (ablation) | NeuralFLoC · Symbols (2 classes) | ACC: 0.787 -> 0.930 | Improved |
| Add clustering loss to registration (ablation) | NeuralFLoC · Shapes | ATV: 24.1 -> 8.1 | Improved |
| Use Fourier basis instead of STFT features (ablation) | NeuralFLoC · Symbols (2 classes) | ACC: 0.886 -> 0.930 | Improved |
| Use Fourier basis instead of STFT features (ablation) | NeuralFLoC · Wave (d=1) | ACC: 0.988 -> 0.993 | No clear effect |
| Joint classical registration+clustering vs NeuralFLoC | BSRC/SRC/JPCCA · all datasets | ACC | Hurt |
| Increase loss weight alpha (ablation) | NeuralFLoC · five datasets | NMI/ATV | No clear effect |
| Increase number of Fourier basis functions K beyond 10 (ablation) | NeuralFLoC · five datasets | NMI/ATV | Mixed |
| Scale up to ~70k samples | NeuralFLoC · Symbols (2 classes) augmented | ACC: 0.930 -> 0.942 | No clear effect |
| Add missing data (5-20%) or irregular sampling | NeuralFLoC · five datasets | NMI/ATV | No clear effect |
Closest papers by meaning
- DeepFRC: An End-to-End Deep Learning Model for Functional Registration and Classification (ICLR 2026 · similarity 0.89)
- Shape-Informed Clustering of Multi-Dimensional Functional Data via Deep Functional Autoencoders (NeurIPS 2025 · similarity 0.85)
- Riemannian Flow Matching for Brain Connectivity Matrices via Pullback Geometry (NeurIPS 2025 · similarity 0.84)
- Latent Functional Maps: a spectral framework for representation alignment (NeurIPS 2024 · similarity 0.83)
- Improving Function Space Flow Matching with Kernel Optimal Transport (NeurIPS 2026 · similarity 0.83)
- Memory-Scalable and Simplified Functional Map Learning (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.
