Research map / Architectures and efficiency / RNNs and normalization
Content-Aware Frequency Encoding for Implicit Neural Representations with Fourier-Chebyshev Features
Junbo Ke, Yangyang Xu, Chao Wang, You-Wei Wen
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 CAFE parallel linear layers with Hadamard product to RFF (ablation) | RFF MLP · 2D image fitting D2K0 | PSNR (dB): 26.18 -> 34.87 | Improved |
| add Chebyshev features to RFF (ablation) | RFF MLP · 2D image fitting D2K0 | PSNR (dB): 26.18 -> 33.00 | Improved |
| add Chebyshev features to CAFE (ablation) | CAFE · 2D image fitting D2K0 | PSNR (dB): 34.87 -> 39.47 | Improved |
| add Chebyshev features to CAFE (ablation) | CAFE · 2D image fitting D2K1 | PSNR (dB): 37.36 -> 44.34 | Improved |
| add CAFE to Chebyshev+Fourier features (ablation) | Fourier-Chebyshev MLP · 2D image fitting D2K2 | PSNR (dB): 35.60 -> 44.18 | Improved |
| increase number of parallel linear layers (ablation) | CAFE+ · 2D image fitting (DIV1) | PSNR (dB): steady gain then saturation (figure only) | Improved |
| remove backbone MLP (0 layers vs 2) (ablation) | CAFE+ · 2D image fitting D2K5-D2K7 | Avg PSNR (dB): 42.26 -> 35.56 | Hurt |
| increase backbone MLP depth beyond 2 layers (ablation) | CAFE+ · 2D image fitting D2K5-D2K7 | Avg PSNR (dB): 42.26 -> 42.17 (3), 42.52 (4) | No clear effect |
| increase depth of plain RFF MLP instead of encoding layers (ablation) | RFF · image fitting | PSNR (dB) | No clear effect |
| CAFE+ vs FINER on Drums | CAFE+ · NeRF Blender Drums | PSNR (dB): 25.66 -> 25.59 | No clear effect |
| CAFE+ vs FINER on Ship | CAFE+ · NeRF Blender Ship | PSNR (dB): 22.02 -> 23.12 | Improved |
| CAFE+ default vs SL2A on D2K5 | CAFE+ (default) · 2D image fitting D2K5 | PSNR (dB): 37.03 -> 36.64 | Hurt |
Closest papers by meaning
- FreSh: Frequency Shifting for Accelerated Neural Representation Learning (ICLR 2025 · similarity 0.89)
- Representing Sounds as Neural Amplitude Fields: A Benchmark of Coordinate-MLPs and a Fourier Kolmogorov-Arnold Framework (AAAI 2025 · similarity 0.87)
- A sampling theory perspective on activations for implicit neural representations (ICML 2024 · similarity 0.87)
- Learning Spatially Collaged Fourier Bases for Implicit Neural Representation (AAAI 2024 · similarity 0.87)
- Tuning the Frequencies: Robust Training for Sinusoidal Neural Networks (CVPR 2025 · similarity 0.86)
- FINER: Flexible Spectral-bias Tuning in Implicit NEural Representation by Variable-periodic Activation Functions (CVPR 2024 · similarity 0.86)
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.
