AI Paper Atlas

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

A simple neural network module for relational reasoning

NeurIPS 2017 · Vision-language and multimodal › Visual question answering · approach: Open-domain multimodal QA

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 changedWhereMeasuredEffect
Add a Relation Network to the CNN+LSTM visual QA architectureCNN+LSTM · CLEVR pixel testoverall accuracy: 68.5 to 95.5Helped
Use a Relation Network instead of an MLP reasoning moduleCNN-based Sort-of-CLEVR model · Sort-of-CLEVR relational questionsaccuracy: 63% to above 94%Helped
Use a Relation Network instead of an MLP reasoning moduleCNN-based Sort-of-CLEVR model · Sort-of-CLEVR non-relational questionsaccuracyNo clear effect
Use a Relation Network for dynamic ball connection inferenceDynamic physical systems model · Connection inferencecorrectly classified test scenesHelped
Use a Relation Network for counting connected systemsDynamic physical systems model · Countingscenes with correct countHelped
Use an RN architecture for basic inductionbAbI · Basic induction tasktotal error: 2.1% total error; Sparse DNC 54%; DNC 55.1%; EntNet 52.1%Helped

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