Research map / Architectures and efficiency / CNN design
Deep Network Classification by Scattering and Homotopy Dictionary Learning
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 |
|---|---|---|---|
| Add a learned positive sparse-code stage to scattering features, using W=D. | Scattering representation with an MLP classifier · ImageNet 2012 | Top5 accuracy: 65.3 -> 81.0 | Helped |
| Add a learned positive sparse-code stage to scattering features, using W=D. | Scattering representation with an MLP classifier · ImageNet 2012 | Top1 accuracy: 42.0 -> 59.2 | Helped |
| Learn W independently instead of constraining W=D. (ablation) | Scattering + ISTC sparse-code classifier · ImageNet 2012 | Top5 accuracy: 81.0 -> 83.7 | Helped |
| Learn W independently instead of constraining W=D. (ablation) | Scattering + ISTC sparse-code classifier · ImageNet 2012 | Top1 accuracy: 59.2 -> 62.8 | Helped |
| Learn W independently instead of constraining W=D. (ablation) | ISTC sparse coding · ImageNet 2012 sparse-code convergence | L1 Lagrangian minimization | Hurt |
| Classify the reconstruction D alpha instead of the sparse code alpha. (ablation) | Scattering + ISTC with W=D · ImageNet 2012 | Top5 accuracy: 81.0 -> 79.3 | Hurt |
| Classify the reconstruction D alpha instead of the sparse code alpha. (ablation) | Scattering + ISTC with W=D · ImageNet 2012 | Top1 accuracy: 59.2 -> 56.9 | Hurt |
| Use the ISTC sparse code with W=D rather than scattering features alone. | Linear classifier · ImageNet 2012 | Top1 accuracy: 26.1% -> 51.6% | Helped |
| Use the ISTC sparse code with W=D rather than scattering features alone. | Linear classifier · ImageNet 2012 | Top5 accuracy: 44.7% -> 73.7% | Helped |
| Use ISTC with W=D instead of ISTA for 12 iterations. | Sparse-code solver · ImageNet 2012 learned dictionary | Relative MSE: 0.45 -> 0.02 | Helped |
| Use ISTC with W=D instead of FISTA for 12 iterations. | Sparse-code solver · ImageNet 2012 learned dictionary | Relative MSE: 0.23 -> 0.02 | Helped |
| Replace the learned dimension-reduction operator L with unsupervised PCA. (ablation) | Sparse scattering network · ImageNet 2012 | classification accuracy | Hurt |
Closest papers by meaning
- Separation and Concentration in Deep Networks (ICLR 2021 · similarity 0.88)
- From Flat to Hierarchical: Extracting Sparse Representations with Matching Pursuit (NeurIPS 2025 · similarity 0.84)
- Unsupervised Parameter-free Simplicial Representation Learning with Scattering Transforms (ICML 2024 · similarity 0.84)
- How Deep Networks Learn Sparse and Hierarchical Data: the Sparse Random Hierarchy Model (ICML 2024 · similarity 0.83)
- HUNet: Homotopy Unfolding Network for Image Compressive Sensing (CVPR 2025 · similarity 0.83)
- Random Sparse Lifts: Construction, Analysis and Convergence of finite sparse 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.