Research map / Vision-language and multimodal / Human and embodied modeling
Right for the Right Concept: Revising Neuro-Symbolic Concepts by Interacting With Their Explanations
Check 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 semantic explanation feedback | NeSy model · CLEVR-Hans3 test (non-confounded) | balanced accuracy: 81.71 ± 3.09 -> 91.31 ± 3.13 | Helped |
| Add semantic explanation feedback | NeSy model · CLEVR-Hans7 test (non-confounded) | balanced accuracy: 90.97 ± 0.91 -> 94.96 ± 0.49 | Helped |
| Add visual XIL with segmentation-mask feedback | CNN · CLEVR-Hans3 test (non-confounded) | balanced accuracy: 70.34 ± 0.30 -> 70.77 ± 0.37 | No clear effect |
| Add visual XIL with segmentation-mask feedback | CNN · CLEVR-Hans7 test (non-confounded) | balanced accuracy: 84.50 ± 1.04 -> 89.26 ± 0.29 | Helped |
| Add a global rule excluding gray from predictions | NeSy model · CLEVR-Hans3 test (class 1) | balanced accuracy: 52.98 ± 9.60 -> 83.59 ± 8.44 | Helped |
| Add a global rule excluding gray from predictions | NeSy model · CLEVR-Hans3 test (all classes) | balanced accuracy: 81.71 ± 3.09 -> 83.26 ± 6.46 | Helped |
| Add semantic explanation feedback | NeSy model · CLEVR-Hans3 test | relative L1 error: 73% reduction (total) | Helped |
| Add semantic explanation feedback | NeSy model · CLEVR-Hans7 test | relative L1 error: 54% reduction (total) | Helped |
Closest papers by meaning
- Knowledge-Aware Neuron Interpretation for Scene Classification (AAAI 2024 · similarity 0.87)
- CoE: Chain-of-Explanation via Automatic Visual Concept Circuit Description and Polysemanticity Quantification (CVPR 2025 · similarity 0.87)
- NeuronEye: Query-Guided Visual Concept Activation for Vision-Language Reasoning (NeurIPS 2026 · similarity 0.87)
- Concept-RuleNet: Grounded Multi-Agent Neurosymbolic Reasoning in Vision Language Models (AAAI 2026 · similarity 0.87)
- Automated Natural Language Explanation of Deep Visual Neurons with Large Models (Student Abstract) (AAAI 2024 · similarity 0.86)
- Language Model as Visual Explainer (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.