Research map / Label-efficient and robust learning / Recognition and long-tail
D-Separation for Causal Self-Explanation
Wei Liu, Jun Wang, Haozhao Wang, Ruixuan Li, Zhiying Deng, YuanKai Zhang, Yang Qiu
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 |
|---|---|---|---|
| use MCD instead of FR | FR · correlated BeerAdvocate Aroma S~10 | F1: 52.1 -> 65.8 | Improved |
| use MCD instead of FR | FR · correlated BeerAdvocate Appearance S~10 | F1: 52.6 -> 63.9 | Improved |
| use MCD instead of FR | FR · correlated BeerAdvocate Palate S~10 | F1: 42.3 -> 53.1 | Improved |
| use MCD instead of FR | FR · correlated BeerAdvocate Appearance S~30 | F1: 62.3 -> 73.4 | Improved |
| use MCD instead of FR | FR · correlated BeerAdvocate Appearance S~20 | F1: 79.6 -> 82.3 | Improved |
| use MCD instead of FR | FR · HotelReview | F1: 44.8 -> 47.8 | Improved |
| replace KL-divergence with JS-divergence in MCD criterion (ablation) | MCD · correlated BeerAdvocate Appearance S~10 | F1: 63.9 -> 65.9 | Improved |
| replace KL-divergence with JS-divergence (ablation) | MCD · correlated BeerAdvocate Palate S~30 | F1: 42.9 -> 37.1 | Hurt |
| replace KL-divergence with JS-divergence (ablation) | MCD · correlated BeerAdvocate Aroma S~20 | F1: 72.8 -> 66.2 | No clear effect |
| increase mask correlation (skew 65 to 75) (ablation) | FR · Palate decorrelated | F1: 63.4 -> 51.0 | Hurt |
| increase mask correlation (skew 65 to 75) (ablation) | MCD · Palate decorrelated | F1: 62.6 -> 63.1 | No clear effect |
| swap GRU encoder for pretrained ELECTRA (ablation) | MCD · decorrelated Beer-Appearance | F1: 80.1 -> 85.2 | Improved |
Closest papers by meaning
- Learning Robust Rationales for Model Explainability: A Guidance-Based Approach (AAAI 2024 · similarity 0.87)
- Is the MMI Criterion Necessary for Interpretability? Degenerating Non-causal Features to Plain Noise for Self-Rationalization (NeurIPS 2024 · similarity 0.87)
- Can Large Language Models Infer Causation from Correlation? (ICLR 2024 · similarity 0.85)
- Breaking Free from MMI: A New Frontier in Rationalization by Probing Input Utilization (ICLR 2025 · similarity 0.85)
- Rethinking the Development of Large Language Models from the Causal Perspective: A Legal Text Prediction Case Study (AAAI 2024 · similarity 0.84)
- Tailoring Self-Rationalizers with Multi-Reward Distillation (ICLR 2024 · similarity 0.84)
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.
