Research map / Image generation / Autoregressive and latent generation
Eigen-Distortions of Hierarchical Representations
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 the early VGG16 layers, especially Front and Layer3, rather than deeper layers | VGG16 · Human detection of eigen-distortions | predictive accuracy for human sensitivity | Helped |
| Use deeper VGG16 layers rather than early layers | VGG16 · Human detection of eigen-distortions | predictive accuracy for human sensitivity | Hurt |
| Add the L2 pooling modification (ablation) | VGG16 · Human detection of eigen-distortions | visibility of eigen-distortions | No clear effect |
| Use the biologically structured On-Off model instead of the generic 4-layer CNN | IQA-optimized models · Human detection of eigen-distortions | predictive accuracy for human sensitivity | Helped |
| Use the biologically structured On-Off model instead of VGG-IQA | IQA-optimized models · Human detection of eigen-distortions | predictive accuracy for human sensitivity | Helped |
| Add LGN functional elements, moving from the more reduced variants toward LGG or On-Off (ablation) | LGN models · TID-2008 held-out test set | Pearson correlation: .66 (LN), .74 (LG), .83 (LGG); On-Off: .82 | Helped |
| Add LGN functional elements, moving from reduced variants toward On-Off (ablation) | LGN models · Human detection of eigen-distortions | predictive accuracy for human sensitivity | Helped |
| Use a model incorporating local gain control rather than the generic CNN or VGG-IQA (ablation) | LGN models · Human detection of eigen-distortions | predictive accuracy for human sensitivity | Helped |
| Use a model incorporating local gain control rather than any VGG16 layer (ablation) | LGN models · Human detection of eigen-distortions | predictive accuracy for human sensitivity | Helped |
Closest papers by meaning
- Discriminating image representations with principal distortions (ICLR 2025 · similarity 0.92)
- Strong and Precise Modulation of Human Percepts via Robustified ANNs (NeurIPS 2023 · similarity 0.87)
- The Unreasonable Effectiveness of Linear Prediction as a Perceptual Metric (ICLR 2024 · similarity 0.86)
- Multidimensional Observer Model and Perceptual Dimensions of Human Image Quality Assessment (NeurIPS 2026 · similarity 0.85)
- L-WISE: Boosting Human Visual Category Learning Through Model-Based Image Selection and Enhancement (ICLR 2025 · similarity 0.85)
- Decomposing stimulus-specific sensory neural information via diffusion models (NeurIPS 2025 · similarity 0.85)
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.