AI Paper Atlas

Research map / Vision-language and multimodal / Cross-modal learning

Audio-Visual Generalised Zero-Shot Learning With Cross-Modal Attention and Language

Otniel-Bogdan Mercea, Lukas Riesch, A. Sophia Koepke, Zeynep Akata

CVPR 2022 · Vision-language and multimodal › Cross-modal learning · approach: Self-supervised audio-visual representation learning · 4 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
Train jointly with audio and visual inputs rather than a single modality (ablation)AVCA · ActivityNet-GZSLHM: 7.53 -> 12.13Improved
Train jointly with audio and visual inputs rather than a single modality (ablation)AVCA · UCF-GZSLHM: 20.92 -> 27.15Improved
Remove cross-attention (ablation)AVCA · ActivityNet-GZSLHM: 12.13 -> 6.50Hurt
Remove audio-branch training components while retaining cross-attention (ablation)AVCA · VGGSound-GZSLHM: 6.31 -> 6.63Mixed
Use audio-branch output rather than visual-branch output for evaluation (ablation)AVCA · ActivityNet-GZSLHM: 12.13 -> 12.54Mixed
Combine audio- and visual-branch outputs at evaluation rather than use the visual output (ablation)AVCA · UCF-GZSLHM: 27.15 -> 25.78Hurt
Remove base triplet loss l_t (ablation)AVCA · ActivityNet-GZSLHM: 12.13 -> 8.39Hurt
Remove base triplet loss l_t (ablation)AVCA · ActivityNet-GZSLZSL: 9.13 -> 9.54Improved
Remove reconstruction loss l_rec (ablation)AVCA · ActivityNet-GZSLHM: 12.13 -> 9.59Hurt
Remove decoded triplet loss l_ct (ablation)AVCA · UCF-GZSLHM: 27.15 -> 17.88Hurt
Remove word-alignment triplet loss l_w (ablation)AVCA · UCF-GZSLHM: 27.15 -> 20.75Hurt
Remove regularisation loss l_r (ablation)AVCA · UCF-GZSLHM: 27.15 -> 21.31Hurt

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