Research map / Probabilistic and scientific modeling / Molecules and proteins
Exploring State-Space Models for Data-Specific Neural Representations
Jinsung Lee, Suha Kwak
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 |
|---|---|---|---|
| replace Mamba/ConvNeXt with S3K convolution in LPNet | LPNet · Kodak | PSNR: 25.99 -> 28.09 (vs ConvNeXt); 27.44 -> 28.09 (vs LPNet-Conv) | Improved |
| use S3K instead of Mamba in LPNet (ablation) | LPNet · Objaverse | PSNR: 17.74 -> 18.34 | Improved |
| use LPNet layout instead of ConvNeXt with convolution (ablation) | ConvNeXt encoder · Kodak | PSNR: 25.99 -> 27.44 | Improved |
| use Laplacian pyramid encoder instead of stacked SSM/conv (ablation) | Mamba encoder · CIFAR-100 reconstruction | PSNR: 24.82 -> 26.58 | Improved |
| stack SSM and conv layers instead of single SSM block (ablation) | S5 encoder · CIFAR-100 reconstruction | PSNR: 25.76 -> 24.84 | Hurt |
| apply SSMs on image pyramid vs single block (ablation) | Mamba encoder · CIFAR-100 reconstruction | PSNR: 24.90 -> 24.78 | Hurt |
| replace Transformer with SSM block (baseline architecture) (ablation) | S4ND encoder · CIFAR-100 reconstruction | PSNR: 24.75 -> 26.00 | Improved |
| add subsequent 1x1 convolution (ablation) | S3K · Bunny (HNeRV decoder) | PSNR: 36.37 -> 36.67 | Improved |
| add adaptive B (ablation) | S3K with 1x1 conv · Bunny (HNeRV decoder) | PSNR: 36.67 -> 36.92 | Improved |
| add adaptive Delta (without inverted bottleneck) (ablation) | S3K with adaptive B · Bunny (HNeRV decoder) | PSNR: 36.92 -> 36.26 | Hurt |
| add SiLU and RMSNorm (ablation) | S3K with inverted bottleneck · Bunny (HNeRV decoder) | PSNR: 36.94 -> 37.03 | Improved |
| replace convolutional NeRV encoder with LPNet-S3K encoder (SNeRV decoder) | SNeRV · UVG | PSNR: 33.28 -> 33.66 | Improved |
Closest papers by meaning
- From Layers to States: A State Space Model Perspective to Deep Neural Network Layer Dynamics (ICLR 2025 · similarity 0.86)
- A scalable generative model for dynamical system reconstruction from neuroimaging data (NeurIPS 2024 · similarity 0.85)
- Spectral State Space Model for Rotation-Invariant Visual Representation Learning (CVPR 2025 · similarity 0.85)
- State Space Models for Event Cameras (CVPR 2024 · similarity 0.84)
- On Feature Learning in Structured State Space Models (NeurIPS 2024 · similarity 0.84)
- GG-SSMs: Graph-Generating State Space Models (CVPR 2025 · similarity 0.84)
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.
