Research map / Architectures and efficiency / RNNs and normalization
Toward the Spectral Bias Alleviation by Normalizations in Coordinate Networks
Zhicheng Cai, Hao Zhu, Qiu Shen, Xinran Wang, Xun Cao
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 cross normalization (CN) | ReLU MLP · Kodak compression [13x49] | PSNR: 24.33 -> 29.22 (up to 4.89dB) | Improved |
| add batch normalization (BN) | ReLU MLP · Kodak compression [13x49] | PSNR: 24.33 -> 29.08 (up to 4.75dB) | Improved |
| add layer normalization (LN) | ReLU MLP · Kodak compression [13x49] | PSNR: 24.33 -> 28.26 (up to 3.93dB) | Improved |
| add global normalization (GN) | ReLU MLP · Kodak compression [13x49] | PSNR: 24.33 -> 27.86 (up to 3.53dB) | Improved |
| add CN to positional encoding network | PE MLP · Kodak compression [13x49] | PSNR: 28.23 -> 31.08 (up to 2.85dB) | Improved |
| add CN to positional encoding network | PE MLP · 3D shape representation | IoU: 0.986 -> 0.997 (average up to 1.12%) | Improved |
| add CN to ReLU network | ReLU MLP · 3D shape representation | IoU: 0.952 -> 0.985 (up to 3.47%) | Improved |
| add CN to ReLU network | ReLU MLP · 3D MRI reconstruction [4x256] | PSNR: 26.11 -> 31.09 (up to 4.98dB) | Improved |
| add CN to PE network | PE MLP · 3D MRI reconstruction [4x256] | PSNR: 32.12 -> 34.97 (up to 2.85dB) | Improved |
| add CN to PE network | PE MLP · 2D CT reconstruction [2x128] | PSNR: 28.16 -> 31.44 | Improved |
| add CN to MipNeRF | MipNeRF · Blender NVS (average) | PSNR: 33.58 -> 33.91 (up to 0.33dB) | Improved |
| add CN to NeuS | NeuS · DTU MVS (15 scans, average) | Chamfer distance: 0.80 -> 0.75 (up to 6.25%) | Improved |
Closest papers by meaning
- Batch Normalization Alleviates the Spectral Bias in Coordinate Networks (CVPR 2024 · similarity 0.96)
- On the Frequency-bias of Coordinate-MLPs (NeurIPS 2022 · similarity 0.87)
- On the Importance of Gaussianizing Representations (ICML 2025 · similarity 0.84)
- Inductive Gradient Adjustment for Spectral Bias in Implicit Neural Representations (ICML 2025 · similarity 0.84)
- Synergistic Integration of Coordinate Network and Tensorial Feature for Improving Neural Radiance Fields from Sparse Inputs (ICML 2024 · similarity 0.84)
- Coordinate-Aware Modulation for Neural Fields (ICLR 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.
