AI Paper Atlas

Research map / Detection and segmentation / CNN features

Complementing Semantic Roles with Temporally Anchored Spatial Knowledge: Crowdsourced Annotations and Experiments

Alakananda Vempala, Eduardo Blanco

AAAI 2016 · Detection and segmentation › CNN features · approach: Local features and context in CNNs · 5 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 machine-learned features instead of most-likely-label baselinespatial knowledge classifier · All instancesoverall F-measure: 0.31 -> 0.52Improved
add lexical features over baseline (ablation)classifier · All instancesoverall F-measure: 0.31 -> 0.48Improved
add syntactic features to lexical features (ablation)classifier · All instancesoverall F-measure: 0.48 -> 0.47No clear effect
add semantic features to lexical + syntactic (ablation)classifier · All instancesoverall F-measure: 0.48 -> 0.52Improved
add semantic and syntactic features on top of lexical (ablation)classifier · Day before instancesF-measure: 0.38 -> 0.41Improved
add semantic and syntactic features on top of lexical (ablation)classifier · Day after instancesF-measure: 0.41 -> 0.48Improved
vary feature set (ablation)classifier · During instancesF-measure: 0.61 -> 0.61No clear effect
restrict evaluation to instances with a majority labelclassifier · All featuresoverall F-measure: 0.52 -> 0.63Improved
generate and filter instances with predicted roles (predicted ∩ gold) instead of gold (ablation)classifier · All featuresoverall F-measure: 0.52 -> 0.55Improved
use fully predicted annotations instead of gold (ablation)classifier · All featuresoverall F-measure: 0.52 -> 0.29Hurt
use fully predicted annotations instead of gold (ablation)classifier · All featuresrecall: 0.54 -> 0.20Hurt
use fully predicted annotations instead of gold (ablation)classifier · All featuresprecision: 0.51 -> 0.58Mixed

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