Research map / Architectures and efficiency / RNNs and normalization
A Learnable Radial Basis Positional Embedding for Coordinate-MLPs
Sameera Ramasinghe, Simon Lucey
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 graph-Laplacian-trained, coordinate-specific super-Gaussian standard deviations instead of matched RFF | Coordinate-MLPs · 1D signal encoding | Test PSNR: 26.03 -> 31.17 | Improved |
| Use graph-Laplacian-trained, coordinate-specific super-Gaussian standard deviations instead of matched RFF | NeRF-style coordinate-MLP · Realistic Synthetic 3D signal encoding | PSNR: 31.13 -> 32.17 | Improved |
| Use graph-Laplacian-trained, coordinate-specific super-Gaussian standard deviations instead of matched RFF | NeRF-style coordinate-MLP · Realistic Synthetic 3D signal encoding | SSIM: 0.947 -> 0.981 | Improved |
| Train embedding parameters end-to-end with network weights instead of using the proposed procedure (ablation) | Super-Gaussian coordinate-MLP · 1D signal encoding | Test PSNR: 31.17 -> 21.44 | Hurt |
| Use uniform standard deviations instead of graph-Laplacian-trained coordinate-specific values (ablation) | Super-Gaussian coordinate-MLP · 1D signal encoding | Test PSNR: 27.19 -> 31.17 | Hurt |
| Use uniform standard deviations instead of graph-Laplacian-trained coordinate-specific values (ablation) | NeRF-style coordinate-MLP · Realistic Synthetic 3D signal encoding | PSNR: 25.41 -> 32.17 | Hurt |
| Use uniform standard deviations instead of graph-Laplacian-trained coordinate-specific values (ablation) | NeRF-style coordinate-MLP · Realistic Synthetic 3D signal encoding | SSIM: 0.891 -> 0.981 | Hurt |
| Use uniform standard deviations instead of graph-Laplacian-trained coordinate-specific values (ablation) | Coordinate-MLP · 2D image encoding, 25% regular training data | mean PSNR: 17.40 -> 19.99 | Hurt |
| Use uniform standard deviations instead of graph-Laplacian-trained coordinate-specific values (ablation) | Coordinate-MLP · 2D image encoding, 10% regular training data | mean PSNR: 16.17 -> 18.68 | Hurt |
| Use uniform standard deviations instead of graph-Laplacian-trained coordinate-specific values (ablation) | Coordinate-MLP · 2D image encoding, 25% random training data | mean PSNR: 16.80 -> 19.33 | Hurt |
| Use uniform standard deviations instead of graph-Laplacian-trained coordinate-specific values (ablation) | Coordinate-MLP · 2D image encoding, 10% random training data | mean PSNR: 14.91 -> 17.64 | Hurt |
| Use one signal's selected RFF parameters across signals rather than tuning per signal (ablation) | RFF coordinate-MLP · 1D signal encoding | Test PSNR: 26.03 -> 23.24 | Hurt |
Closest papers by meaning
- Trading Positional Complexity vs Deepness in Coordinate Networks (ECCV 2022 · similarity 0.91)
- Learning Efficient Positional Encodings with Graph Neural Networks (ICLR 2025 · similarity 0.84)
- Coordinate-Aware Modulation for Neural Fields (ICLR 2024 · similarity 0.84)
- Learnable Spatial-Temporal Positional Encoding for Link Prediction (ICML 2025 · similarity 0.83)
- Maximizing the Position Embedding for Vision Transformers with Global Average Pooling (AAAI 2025 · similarity 0.83)
- Geographic Location Encoding with Spherical Harmonics and Sinusoidal Representation Networks (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.
