Research map / Vision-language and multimodal / Human and embodied modeling
Learn-Explain-Reinforce: Counterfactual Reasoning and its Guidance to Reinforce an Alzheimer's Disease Diagnosis Model
Kwanseok Oh, Jee Seok Yoon, Heung‐Il Suk
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 |
|---|---|---|---|
| fine-tune backbone with counterfactual-map augmented images | ResNet18 · ADNI CN vs MCI vs AD | accuracy: 0.5802 -> 0.5883 | Improved |
| fine-tune backbone with counterfactual-map augmented images | ResNet18 · ADNI CN vs AD | accuracy: 0.8898 -> 0.9231 | Improved |
| add XGA guided by counterfactual maps instead of augmentation (ablation) | ResNet18 · ADNI CN vs MCI vs AD | accuracy: 0.5883 -> 0.6715 | Improved |
| add XGA guided by counterfactual maps instead of augmentation (ablation) | ResNet18 · ADNI CN vs AD | accuracy: 0.9231 -> 0.9489 | Improved |
| add XGA guided by counterfactual maps | ResNet18 · ADNI sMCI vs pMCI | accuracy: 0.6946 -> 0.7703 | Improved |
| add XGA guided by counterfactual maps | ResNet18 · ADNI CN vs MCI vs AD | mAUC: 0.7501 -> 0.8123 | Improved |
| add second iteration of explanation-reinforcement learning (ablation) | LEAR ResNet18 · ADNI CN vs MCI vs AD | accuracy: 0.6347 -> 0.6715 | Improved |
| add third iteration of explanation-reinforcement learning (ablation) | LEAR ResNet18 · ADNI CN vs MCI vs AD | accuracy: 0.6715 -> 0.6715 | No clear effect |
| add third iteration of explanation-reinforcement learning (ablation) | LEAR ResNet18 · ADNI CN vs AD | accuracy: 0.9515 -> 0.9489 | Hurt |
| add second iteration (ablation) | LEAR ResNet18 · ADNI MCI vs AD | accuracy: 0.8396 -> 0.8716 | Improved |
| add explanation-guided attention to the diagnostic model, comparing to VA-GAN maps | CMG · ADNI CN<->AD NCC(+) | NCC: 0.317 -> 0.366 | Improved |
| use CMG counterfactual maps instead of Grad-CAM | explanation method · ADNI CN<->AD NCC(+) | NCC: 0.227 -> 0.366 | Improved |
Closest papers by meaning
- EMAD: Evidence-Centric Grounded Multimodal Diagnosis for Alzheimer's Disease (CVPR 2026 · similarity 0.86)
- LeapFactual: Reliable Visual Counterfactual Explanation Using Conditional Flow Matching (NeurIPS 2025 · similarity 0.85)
- Counterfactual Explanations for Misclassified Images: How Human and Machine Explanations Differ (Abstract Reprint) (AAAI 2024 · similarity 0.83)
- Rethinking Visual Counterfactual Explanations Through Region Constraint (ICLR 2025 · similarity 0.83)
- Structure Your Data: Towards Semantic Graph Counterfactuals (ICML 2024 · similarity 0.83)
- V-CECE: Visual Counterfactual Explanations via Conceptual Edits (NeurIPS 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.
