Research map / Architectures and efficiency / RNNs and normalization
Polynomial Implicit Neural Representations for Large Diverse Datasets
Read the paperCheck your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (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 polynomial coordinate representations with progressively increased polynomial order | Poly-INR · ImageNet 128 × 128 | FID: 1.81 -> 2.08 | Hurt |
| Use polynomial coordinate representations with progressively increased polynomial order | Poly-INR · ImageNet 256 × 256 | FID: 2.30 -> 2.86 | Hurt |
| Use polynomial coordinate representations with progressively increased polynomial order | Poly-INR · ImageNet 512 × 512 | FID: 2.41 -> 3.81 | Hurt |
| Use a fixed 46.0M-parameter generator across resolutions | Poly-INR · ImageNet 512 × 512 | parameters (M): 168.4 -> 46.0 | Helped |
| Use polynomial INR instead of CIPS | Poly-INR · FFHQ 256 × 256 | FID: 4.38 -> 2.72 | Helped |
| Use polynomial INR instead of INR-GAN | Poly-INR · FFHQ 256 × 256 | FID: 4.95 -> 2.72 | Helped |
| Use polynomial INR instead of StyleGAN-XL | Poly-INR · FFHQ 256 × 256 | FID: 2.19 -> 2.72 | Hurt |
| Use polynomial INR instead of StyleGAN-XL | Poly-INR · FFHQ 256 × 256 | parameters (M): 67.9 -> 13.6 | Helped |
| Generate 512 × 512 images with Poly-INR rather than bicubic upsampling | Poly-INR trained at 32×32 · ImageNet, evaluated at 512 × 512 | FID: 73.86 -> 65.15 | Helped |
| Generate 1024×1024 samples with Poly-INR rather than INR-GAN | Poly-INR trained on FFHQ 256 × 256 · FFHQ, evaluated at 1024×1024 | FID: 18.51 -> 13.69 | Helped |
| Generate 1024×1024 samples with Poly-INR rather than CIPS | Poly-INR trained on FFHQ 256 × 256 · FFHQ, evaluated at 1024×1024 | FID: 29.59 -> 13.69 | Helped |
| Optimize affine parameters for image inversion | Poly-INR · ImageNet validation set | PSNR: 13.5 -> 26.52 | Helped |
Closest papers by meaning
- Implicit Representations via Operator Learning (ICML 2024 · similarity 0.90)
- Grids Often Outperform Implicit Neural Representation at Compressing Dense Signals (NeurIPS 2025 · similarity 0.89)
- VI^3NR: Variance Informed Initialization for Implicit Neural Representations (CVPR 2025 · similarity 0.88)
- ReLUs Are Sufficient for Learning Implicit Neural Representations (ICML 2024 · similarity 0.87)
- MIRE: Matched Implicit Neural Representations (CVPR 2025 · similarity 0.87)
- Learning Input Encodings for Kernel-Optimal Implicit Neural Representations (ICML 2025 · similarity 0.87)
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.