Research map / Detection and segmentation / Open-set and few-shot detection
Strong and Precise Modulation of Human Percepts via Robustified ANNs
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 robustified guide model (eps_tr 3.0) instead of vanilla | guide model · Disruption Modulation, budget 30 | human disruption rate: 90% | Helped |
| robust training budget 1.0 instead of 3.0 (ablation) | guide model · DM | human-model alignment | Hurt |
| robust training budget 10.0 instead of 3.0 (ablation) | guide model · DM | disruption per budget level | Hurt |
| use robustified (3.0/10.0) guide instead of vanilla | guide model · Targeted Modulation, budget 30 | target choice probability: 60% | Helped |
| use image interpolation instead of robust guide | baseline · TM | target choice probability | No clear effect |
| use vanilla model as TM guide | guide model · TM | target choice probability | No clear effect |
| change source images from in-distribution to OOD | Robust eps_tr 3.0 · TM | normalized induction rate: 60% -> 51% | Hurt |
| change source images from in-distribution to ANI | Robust eps_tr 3.0 · TM | normalized induction rate: 60% -> 55% | Hurt |
| change source images from in-distribution to uniform noise | Robust eps_tr 3.0 · TM | normalized induction rate: 60% -> 70% | Helped |
| robust training budget 3.0 vs 10.0 vs 1.0 on TM efficiency (ablation) | Robust model · TM MTME | budget to reach 10% (lower better): 9 vs 11 vs 196 | Mixed |
| use l-infinity instead of l2 constraint for TM (ablation) | robust model · TM | modulation strength | Hurt |
| vary viewing time 100-800ms | human experiment · DM | disruption | No clear effect |
Closest papers by meaning
- L-WISE: Boosting Human Visual Category Learning Through Model-Based Image Selection and Enhancement (ICLR 2025 · similarity 0.88)
- Testing the Limits of Fine-Tuning for Improving Visual Cognition in Vision Language Models (ICML 2025 · similarity 0.87)
- Synthesizing Images on Perceptual Boundaries of ANNs for Uncovering and Manipulating Human Perceptual Variability (ICML 2025 · similarity 0.87)
- Directional Confusions Reveal Divergent Inductive Biases Through Rate-Distortion Geometry in Human and Machine Vision (NeurIPS 2026 · similarity 0.87)
- Not Too Generative, Not Too Discriminative: The Human Alignment Sweet Spot (NeurIPS 2026 · similarity 0.87)
- Evaluating alignment between humans and neural network representations in image-based learning tasks (NeurIPS 2024 · similarity 0.86)
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.