Research map / Vision-language and multimodal / Cross-modal learning
Temporal and Cross-Modal Attention for Audio-Visual Zero-Shot Learning
Otniel-Bogdan Mercea, Thomas Hummel, A. Sophia Koepke, Zeynep Akata
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 changed | Where | Measured | Effect |
|---|---|---|---|
| add cross-entropy loss to regression loss (ablation) | TCaF · VGGSound-GZSL | ZSL accuracy: 2.50 -> 4.31 | Improved |
| add reconstruction loss to regression and cross-entropy (ablation) | TCaF · UCF-GZSL | GZSL harmonic mean (HM): 49.76 -> 50.78 | Improved |
| add cross-entropy loss to regression loss (ablation) | TCaF · UCF-GZSL | GZSL harmonic mean (HM): 16.25 -> 49.76 | Improved |
| add temporal embeddings (ablation) | TCaF · ActivityNet-GZSL | GZSL harmonic mean (HM): 8.69 -> 12.20 | Improved |
| use cross-attention instead of full attention with class token (ablation) | TCaF · UCF-GZSL | GZSL harmonic mean (HM): 39.18 -> 50.78 | Improved |
| use cross-attention instead of self-attention with class token (ablation) | TCaF · VGGSound-GZSL | GZSL harmonic mean (HM): 6.71 -> 8.77 | Improved |
| add classification token (ablation) | TCaF · VGGSound-GZSL | GZSL harmonic mean (HM): 6.27 -> 8.77 | Improved |
| add audio modality to visual-only input (ablation) | TCaF · UCF-GZSL | GZSL harmonic mean (HM): 31.49 -> 50.78 | Improved |
| add visual modality to audio-only input (ablation) | TCaF · UCF-GZSL | GZSL harmonic mean (HM): 25.38 -> 50.78 | Improved |
| use temporal sequence instead of temporally averaged inputs (ablation) | TCaF · VGGSound-GZSL | GZSL harmonic mean (HM): 7.65 -> 8.77 | Improved |
| replace training objective with triplet losses (ablation) | TCaF · VGGSound-GZSL | GZSL harmonic mean (HM): 8.77 -> 7.19 | Hurt |
| add feed forward functions to cross-attention layers (ablation) | TCaF · UCF-GZSL | GZSL harmonic mean (HM): 28.78 -> 50.78 | Improved |
Closest papers by meaning
- Audio-Visual Generalised Zero-Shot Learning With Cross-Modal Attention and Language (CVPR 2022 · similarity 0.95)
- Few-Shot Audio-Visual Class-Incremental Learning with Temporal Prompting and Regularization (AAAI 2025 · similarity 0.87)
- An eye for an ear: zero-shot audio description leveraging an image captioner with audio-visual token distribution matching (NeurIPS 2024 · similarity 0.87)
- CAV-MAE Sync: Improving Contrastive Audio-Visual Mask Autoencoders via Fine-Grained Alignment (CVPR 2025 · similarity 0.87)
- Visual and Semantic Prompt Collaboration for Generalized Zero-Shot Learning (CVPR 2025 · similarity 0.86)
- CrossMAE: Cross-Modality Masked Autoencoders for Region-Aware Audio-Visual Pre-Training (CVPR 2024 · similarity 0.86)
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.
