Research map / Vision-language and multimodal / Visual question answering
A Semi-Supervised Learning Approach to Why-Question Answering
Jong-Hoon Oh, Kentaro Torisawa, Chikara Hashimoto, Ryu Iida, Masahiro Tanaka, Julien Kloetzer
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 |
|---|---|---|---|
| semi-supervised iterative training with comprehensible-question filtered auto-labeled data | why-QA answer ranker (INIT) · Japanese why-QA test set | P@1: 42.0 -> 50.0 | Improved |
| semi-supervised iterative training | why-QA answer ranker (INIT) · Japanese why-QA test set | MAP: 46.5 -> 48.9 | Improved |
| same semi-supervised method on own answer retrieval | why-QA answer ranker · Own answer retrieval | P@1: 43.0 -> 50.0 | Improved |
| add comprehensible-question filtering (ablation) | OURS(USC) · Japanese why-QA test set | P@1: 34.0 -> 50.0 | Improved |
| add comprehensible-question filtering (ablation) | OURS(USC) · Japanese why-QA test set | MAP: 41.7 -> 48.9 | Improved |
| exclude passages containing the source causal relation (ablation) | OURS(USC) · Japanese why-QA test set | P@1: 42.0 -> 50.0 | Improved |
| exclude passages containing the source causal relation (ablation) | OURS(USC) · Japanese why-QA test set | MAP: 44.4 -> 48.9 | Improved |
| iterate with top-K uncertain examples instead of adding all labeled examples at once (ablation) | OURS(USC) · Japanese why-QA test set | P@1: 42.0 -> 50.0 | Improved |
| iterate with top-K uncertain examples instead of adding all at once (ablation) | OURS(USC) · Japanese why-QA test set | MAP: 45.4 -> 48.9 | Improved |
| naively use causal relation source text as training answers | why-QA ranker · preliminary experiments | performance | Hurt |
Closest papers by meaning
- Semi-Distantly Supervised Neural Model for Generating Compact Answers to Open-Domain Why Questions (AAAI 2018 · similarity 0.89)
- LLCP: Learning Latent Causal Processes for Reasoning-based Video Question Answer (ICLR 2024 · similarity 0.84)
- QASA: Advanced Question Answering on Scientific Articles (ICML 2023 · similarity 0.84)
- Learning From Unannotated QA Pairs to Analogically Disambiguate and Answer Questions (AAAI 2018 · similarity 0.83)
- CausalChaos! Dataset for Comprehensive Causal Action Question Answering Over Longer Causal Chains Grounded in Dynamic Visual Scenes (NeurIPS 2024 · similarity 0.82)
- Causal-CoG: A Causal-Effect Look at Context Generation for Boosting Multi-modal Language Models (CVPR 2024 · similarity 0.82)
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.
