Research map / Vision-language and multimodal / Visual question answering
Two Can Play This Game: Visual Dialog With Discriminative Question Generation and Answering
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 |
|---|---|---|---|
| jointly score each answer option with fused query, image, caption, and history features | VisDial answer prediction · VisDial v0.9 test | Recall@5: 78.75 -> 78.96 | Helped |
| add image context to the query-only model (ablation) | SF visual dialog model · VisDial v0.9 test | Recall@5: 72.75 -> 76.73 | Helped |
| add caption and history to the query-and-image model (ablation) | SF visual dialog model · VisDial v0.9 test | Recall@5: 76.73 -> 78.96 | Helped |
| use two hidden MLP layers instead of one (ablation) | SF visual dialog model · VisDial v0.9 test | Recall@5: 78.66 -> 78.96 | Helped |
| use shared language embedding matrices across LSTMs (ablation) | SF visual dialog model · VisDial v0.9 test | Recall@5: 77.69 -> 78.66 | Helped |
| add image context to the query-only model (ablation) | SF visual question generation model · VisDial-Q test | Recall@5: 26.44 -> 42.32 | Helped |
| add caption and history to the query-and-image model (ablation) | SF visual question generation model · VisDial-Q test | Recall@5: 42.32 -> 55.17 | Helped |
| use two hidden MLP layers instead of one (ablation) | SF visual question generation model · VisDial-Q test | Recall@5: 54.74 -> 55.17 | Helped |
| use shared language embedding matrices across LSTMs (ablation) | SF visual question generation model · VisDial-Q test | Recall@5: 53.03 -> 54.74 | Helped |
Closest papers by meaning
- Visual Dialog (CVPR 2017 · similarity 0.94)
- Image-Question-Answer Synergistic Network for Visual Dialog (CVPR 2019 · similarity 0.92)
- Best of Both Worlds: Transferring Knowledge from Discriminative Learning to a Generative Visual Dialog Model (NeurIPS 2017 · similarity 0.91)
- UTC: A Unified Transformer With Inter-Task Contrastive Learning for Visual Dialog (CVPR 2022 · similarity 0.91)
- Divide-and-Conquer: Tree-structured Strategy with Answer Distribution Estimator for Goal-Oriented Visual Dialogue (AAAI 2025 · similarity 0.89)
- ConVQG: Contrastive Visual Question Generation with Multimodal Guidance (AAAI 2024 · similarity 0.88)
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.