AI Paper Atlas

Research map / Vision-language and multimodal / Visual question answering

DMRM: A Dual-Channel Multi-Hop Reasoning Model for Visual Dialog

Feilong Chen, Fandong Meng, Jiaming Xu, Peng Li, Bo Xu, Jie Zhou

AAAI 2020 · Vision-language and multimodal › Visual question answering · approach: Vision-language model VQA evaluation · 27 citations (OpenAlex)

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 changedWhereMeasuredEffect
use dual-channel multi-hop reasoning instead of single-hop reasoningDMRM · VisDial val v0.9MRR: 51.99 -> 55.96Improved
use dual-channel reasoning instead of single-channel reasoningDMRM · VisDial val v0.9MRR: 55.43 -> 55.96Improved
add the Locate module (ablation)DMRM · VisDial val v0.9MRR: 54.77 -> 55.96Improved
add the Track module (ablation)DMRM · VisDial val v0.9MRR: 53.28 -> 55.96Improved
add multimodal attention to the decoder (ablation)DMRM · VisDial val v0.9MRR: 55.57 -> 55.96Improved
use 3-hop rather than 1-hop reasoning (ablation)DMRM without multimodal attention decoder · VisDial val v0.9MRR: 55.04 -> 55.57Improved
use 3-hop rather than 2-hop reasoning (ablation)DMRM without multimodal attention decoder · VisDial val v0.9MRR: 54.87 -> 55.57Improved
use 3-hop reasoning rather than the full model's 3-hop configuration (ablation)DMRM without multimodal attention decoder · VisDial val v0.9MRR: 55.57 -> 55.57No clear effect
use DMRM rather than HCIAEvisual dialog response generation · VisDial val v0.9Turing-test pass rate: 0.60 -> 0.65Improved
use DMRM rather than HCIAEvisual dialog response generation · VisDial val v0.9better-or-equal-to-human rate: 0.53 -> 0.62Improved

Closest papers by meaning

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.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy