Research map / Vision-language and multimodal / Visual question answering
Deep Attention Neural Tensor Network for Visual Question Answering
Yalong Bai, Jianlong Fu, Tiejun Zhao, Tao Mei
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 |
|---|---|---|---|
| add neural tensor network (k=6) over image-question-answer triplets | MUTAN · VQA-2.0 val | overall accuracy: 62.84 -> 63.83 | Improved |
| add attention module for reasoning to NTN (k=6) (ablation) | MUTAN + NTN · VQA-2.0 val | overall accuracy: 63.83 -> 64.16 | Improved |
| add attention module for reasoning to NTN (k=3) (ablation) | MUTAN + NTN · VQA-2.0 val | overall accuracy: 63.74 -> 64.07 | Improved |
| add attention module for reasoning to NTN (k=10) (ablation) | MUTAN + NTN · VQA-2.0 val | overall accuracy: 63.86 -> 64.07 | Improved |
| increase number of tensor slices k from 3 to 10 (ablation) | MUTAN + NTN · VQA-2.0 val | overall accuracy: 63.74 -> 63.86 | No clear effect |
| increase number of tensor slices k from 3 to 6 to 10 (ablation) | MUTAN + DA-NTN · VQA-2.0 val | overall accuracy: 64.07 -> 64.16 -> 64.07 | No clear effect |
| add DA-NTN | MLB · VQA-2.0 val | overall accuracy: 62.98 -> 64.58 | Improved |
| add DA-NTN | MUTAN · VQA-2.0 test-dev | overall accuracy: 66.01 -> 67.15 | Improved |
| add DA-NTN | MLB · VQA-2.0 test-dev | overall accuracy: 66.27 -> 67.56 | Improved |
| add DA-NTN | MLB · VQA-2.0 test-standard | overall accuracy: 66.62 -> 67.94 | Improved |
| add DA-NTN | MUTAN (bottom-up) · VQA-1.0 test-dev | overall accuracy: 65.7 -> 67.1 | Improved |
| add DA-NTN | MLB (bottom-up) · VQA-1.0 test-dev | overall accuracy: 65.9 -> 67.9 | Improved |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.
- MUTAN — MUTAN: Multimodal Tucker Fusion for Visual Question Answering (ICCV 2017) · better on 6 of 6 shared results
Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.
Closest papers by meaning
- Multi-Level Attention Networks for Visual Question Answering (CVPR 2017 · similarity 0.93)
- Tips and Tricks for Visual Question Answering: Learnings From the 2017 Challenge (CVPR 2018 · similarity 0.92)
- Deep Modular Co-Attention Networks for Visual Question Answering (CVPR 2019 · similarity 0.92)
- An Analysis of Visual Question Answering Algorithms (ICCV 2017 · similarity 0.92)
- Structured Attentions for Visual Question Answering (ICCV 2017 · similarity 0.92)
- Ask Your Neurons: A Deep Learning Approach to Visual Question Answering (IJCV 2017 · similarity 0.91)
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.
