Research map / Adversarial robustness and security / Certified and empirical robustness
Approximate Feature Collisions in Neural Nets
Ke Li, Tianhao Zhang, Jitendra Malik
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 |
|---|---|---|---|
| Interpolate inside polytope found from target initialization | feature-collision method · MNIST | % classified as target class: 100% | Improved |
| Search over compositions of image patches instead of raw pixels | feature-collision method · ImageNet VGG-16 | % successful collisions: 100% | No clear effect |
| Constrain search to patch compositions (more perceptually different vertices) (ablation) | feature-collision method · ImageNet VGG-16 | top class probability, interpolated samples: 0.995 -> 0.877 | Hurt |
| Search over patch compositions rather than pixels (ablation) | feature-collision method · ImageNet VGG-16 | average distance between polytope vertices: 4906.56 -> 8597.34 | No clear effect |
| Apply method on ImageNet VGG-16 instead of MNIST MLP | feature-collision method · polytope size relative to image distance | vertex distance as fraction of average image distance: 12-13% -> 12-24% | Mixed |
Closest papers by meaning
- Adversarial Examples in Multi-Layer Random ReLU Networks (NeurIPS 2021 · similarity 0.87)
- Decision Boundary Analysis of Adversarial Examples (ICLR 2018 · similarity 0.86)
- Wide Two-Layer Networks can Learn from Adversarial Perturbations (NeurIPS 2024 · similarity 0.85)
- Theoretical Understanding of Learning from Adversarial Perturbations (ICLR 2024 · similarity 0.85)
- Make Haste Slowly: A Theory of Emergent Structured Mixed Selectivity in Feature Learning ReLU Networks (ICLR 2025 · similarity 0.84)
- Feature Averaging: An Implicit Bias of Gradient Descent Leading to Non-Robustness in Neural Networks (ICLR 2025 · 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.
