AI Paper Atlas

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

Interpretable Counting for Visual Question Answering

Alexander Trott, Caiming Xiong, Richard Socher

ICLR 2018 · Vision-language and multimodal › Visual question answering · approach: Universal VQA models · 40 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 sequential discrete object selection with learned object interactions instead of attention-based answer classificationIRLC vs UpDown · HowMany-QA testaccuracy: 52.7 -> 57.7Improved
use sequential discrete object selection with learned object interactions instead of attention-based answer classificationIRLC vs UpDown · HowMany-QA testRMSE: 2.64 -> 2.37Improved
use sequential discrete object selection instead of fractional object countsIRLC vs SoftCount · HowMany-QA testaccuracy: 50.2 -> 57.7Improved
use sequential discrete object selection instead of fractional object countsIRLC vs SoftCount · HowMany-QA testRMSE: 2.37 -> 2.37No clear effect
add caption grounding during training (ablation)IRLC · HowMany-QA testaccuracy: 56.1 -> 57.7Improved
add caption grounding during training (ablation)IRLC · HowMany-QA testRMSE: 2.45 -> 2.37Improved
add caption grounding during training (ablation)UpDown · HowMany-QA testaccuracy: 51.5 -> 52.7Improved
add caption grounding during training (ablation)SoftCount · HowMany-QA testaccuracy: 49.2 -> 50.2Improved
add Visual Genome QA training data (ablation)counting models · HowMany-QA testaccuracy: 2.7%Improved
add Visual Genome QA training data (ablation)counting models · HowMany-QA testRMSE: 0.12Improved
include entropy and interaction auxiliary penalties during training (ablation)IRLC · HowMany-QA developmentaccuracyImproved
increase the weights of entropy and interaction auxiliary penalties too much (ablation)IRLC · HowMany-QA developmentaccuracyHurt

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