Research map / Architectures and efficiency / Attention and transformers
Real-Time Emotion Recognition via Attention Gated Hierarchical Memory Network
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 changed | Where | Measured | Effect |
|---|---|---|---|
| Use a BiGRU rather than a CNN as the utterance reader (ablation) | AGHMN · IEMOCAP | mF1: 59.4 -> 61.9; 59.9 -> 63.0 | Helped |
| Use a BiGRU rather than a BiLSTM as the utterance reader (ablation) | AGHMN · IEMOCAP | mF1: 61.2 -> 61.9; 61.9 -> 63.0 | Helped |
| Replace the BiGRU reader with CNN plus Soft Attention (ablation) | AGHMN · IEMOCAP | mF1: 61.5 -> 58.1 | Hurt |
| Use BiF instead of UniF memory fusion (ablation) | AGHMN with AGRU · IEMOCAP | mF1: 61.5 -> 63.0 | Helped |
| Use BiF instead of UniF memory fusion (ablation) | AGHMN with AGRU · MELD | mF1: 38.0 -> 38.5 | Helped |
| Use BiAGRU instead of AGRU for memory summarization (ablation) | AGHMN with UniF · IEMOCAP | mF1: 61.5 -> 61.9 | Helped |
| Use BiAGRU instead of AGRU for memory summarization (ablation) | AGHMN with UniF · MELD | mF1: 38.0 -> 38.6 | Helped |
| Replace AGRU with Soft Attention (ablation) | AGHMN with UniF · IEMOCAP | mF1: 59.5 -> 61.5 | Hurt |
| Replace AGRU with Soft Attention (ablation) | AGHMN with UniF · MELD | mF1: 37.7 -> 38.0 | Hurt |
| Use BiF memory instead of UniF with Soft Attention (ablation) | AGHMN · IEMOCAP | mF1: 59.5 -> 61.8 | Helped |
| Use BiF memory instead of UniF with Soft Attention (ablation) | AGHMN · MELD | mF1: 37.7 -> 37.5 | No clear effect |
| Combine BiF memory and BiAGRU rather than using either bidirectional component alone (ablation) | AGHMN · IEMOCAP | mF1: 63.0 -> 62.6; 61.9 -> 62.6 | Hurt |
Closest papers by meaning
- Utterance-level Emotion Recognition in Conversation with Conversation-level Supervision (AAAI 2025 · similarity 0.86)
- Dynamic Interactive Bimodal Hypergraph Networks for Emotion Recognition in Conversations (AAAI 2025 · similarity 0.85)
- BIG-FUSION: Brain-Inspired Global-Local Context Fusion Framework for Multimodal Emotion Recognition in Conversations (AAAI 2025 · similarity 0.85)
- Dual-View Learning for Conversational Emotion Recognition Through Context and Emotion-Shift Modeling (AAAI 2025 · similarity 0.85)
- Hardware-aligned Hierarchical Sparse Attention for Efficient Long-term Memory Access (NeurIPS 2025 · similarity 0.84)
- Compressed Context Memory for Online Language Model Interaction (ICLR 2024 · similarity 0.83)
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.