Research map / Label-efficient and robust learning / Recognition and long-tail
Input Similarity from the Neural Network Perspective
Guillaume Charpiat, Nicolas Girard, Loris Felardos, Yuliya Tarabalka
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 |
|---|---|---|---|
| Iterative train-and-realign training for 3 rounds on noisy OSM labels | Registration network · bloomington22 registration | median registration error (pixels): 18 pixels -> 3.5 pixels | Improved |
| Add extra noise to the ground truth and retrain (ablation) | Multi-round registration training · bloomington22 registration | registration accuracy | No clear effect |
| Use extreme thresholds (very high or low) in hard-threshold neighbor estimators (ablation) | Neighbor count estimators · Toy sinusoid frequency problem | neighbor count estimation quality | Hurt |
| Increase sinusoid frequency to be predicted | Neighbor count estimators · Toy sinusoid problem | neighbor count | Hurt |
| Replace perceptual (VGG) loss similarity with network-based similarity for nearest neighbor retrieval | Patch retrieval · Remote sensing patches | qualitative retrieval of similar areas | Improved |
| Later training rounds (round 1 vs 2 vs 3) change neighborhood statistics | Multi-round network · Example patch | soft neighbor count: 1276.7 -> 2486.3 -> 394.6 | Mixed |
Closest papers by meaning
- Going Beyond Neural Network Feature Similarity: The Network Feature Complexity and Its Interpretation Using Category Theory (ICLR 2024 · similarity 0.83)
- Decoupling Semantic Similarity from Spatial Alignment for Neural Networks. (NeurIPS 2024 · similarity 0.82)
- Representational Similarity via Interpretable Visual Concepts (ICLR 2025 · similarity 0.81)
- Not all solutions are created equal: An analytical dissociation of functional and representational similarity in deep linear neural networks (ICML 2025 · similarity 0.80)
- Differentiable Optimization of Similarity Scores Between Models and Brains (ICLR 2025 · similarity 0.80)
- How Not to Stitch Representations to Measure Similarity: Task Loss Matching Versus Direct Matching (AAAI 2025 · similarity 0.80)
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.
