AI Paper Atlas

Research map / Architectures and efficiency / CNN design

Image Question Answering Using Convolutional Neural Network With Dynamic Parameter Prediction

Hyeonwoo Noh, Paul Hongsuck Seo, Bohyung Han

CVPR 2016 · Architectures and efficiency › CNN design · approach: Filter design and trained-filter analysis · 360 citations (OpenAlex)

Check 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 dynamic parameter prediction instead of concatenating image and question features (ablation)DPPnet · VQA test-devaccuracy: 54.70 -> 55.46Improved
pre-train GRU on a large text corpus (ablation)DPPnet · VQA test-devaccuracy: 55.46 -> 56.74Improved
fine-tune CNN weights (ablation)DPPnet · VQA test-devaccuracy: 56.74 -> 57.22Improved
use dynamic parameter prediction instead of concatenating image and question features (ablation)DPPnet · VQA test-devmultiple-choice accuracy: 59.92 -> 61.18Improved
add GRU pre-training (ablation)DPPnet · VQA test-devmultiple-choice accuracy: 61.18 -> 61.95Improved
fine-tune CNN weights (ablation)DPPnet · VQA test-devmultiple-choice accuracy: 61.95 -> 62.48Improved
fine-tune the CNN directly from the start rather than delaying updates to transferred layersDPPnet · ImageQAaccuracyNo clear effect
use an un-pretrained GRU rather than a pre-trained GRU (ablation)DPPnet · VQA test-devaccuracy: 56.74 -> 55.46Hurt

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