Research map / Image generation / GANs
Multi-Task Learning with Generative Adversarial Training for Multi-Passage Machine Reading Comprehension
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 |
|---|---|---|---|
| add generative adversarial training on unlabeled passages (discriminator) to extractor training (ablation) | MG-MRC · Quasar-T test | EM: 46.2 -> 47.1 | Helped |
| train extractor by backpropagation via hybrid method instead of policy gradient (ablation) | MG-MRC · Quasar-T test | EM: 46.5 -> 47.1 | Helped |
| add selector to the extractor (ablation) | MG-MRC · Quasar-T test | EM: 39.0 -> 47.1 | Helped |
| MG-MRC extractor instead of JT-QA extractor (both without selector) | extract-then-select · SearchQA test | EM: 44.7 -> 56.5 | Helped |
| add selection score to final answer scoring (ablation) | MG-MRC · Quasar-T dev | EM: 38.5 -> 46.9 | Helped |
| add boundary score to the product with selection score (ablation) | MG-MRC · Quasar-T dev | EM: 46.6 -> 46.9 | Helped |
| increase candidates per passage K from 1 to 3 (ablation) | MG-MRC · Quasar-T dev | EM: 46.3 -> 46.9 | Helped |
| increase K from 3 to 4 (ablation) | MG-MRC · Quasar-T dev | EM: 46.9 -> 46.7 | No clear effect |
| MG-MRC vs RE3QA (BERT-based, alias labels) | TriviaQA-unfiltered · TriviaQA-unfiltered dev | EM: 65.5 -> 63.4 | Hurt |
Closest papers by meaning
- MMM: Multi-Stage Multi-Task Learning for Multi-Choice Reading Comprehension (AAAI 2020 · similarity 0.89)
- A Robust Adversarial Training Approach to Machine Reading Comprehension (AAAI 2020 · similarity 0.88)
- S-Net: From Answer Extraction to Answer Synthesis for Machine Reading Comprehension (AAAI 2018 · similarity 0.88)
- A Deep Cascade Model for Multi-Document Reading Comprehension (AAAI 2019 · similarity 0.87)
- Multi-Matching Network for Multiple Choice Reading Comprehension (AAAI 2019 · similarity 0.86)
- PMRC: Prompt-Based Machine Reading Comprehension for Few-Shot Named Entity Recognition (AAAI 2024 · 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.