Research map / Label-efficient and robust learning / Recognition and long-tail
Ellipsoid-Based Decision Boundaries for Open Intent Classification
Yuetian Zou, Hanlei Zhang, Hua Xu, Songze Li, Long Xiao
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 |
|---|---|---|---|
| use ellipsoid boundaries instead of ball boundaries with tuned coverage (ablation) | EliDecide · OOS F1 | F1: 83.20 -> 84.48 (KCR 25%, best ball CF=0.975) | Improved |
| add supervised contrastive learning instead of Softmax classifier (ablation) | EliDecide · OOS | F1: 83.78 -> 84.48 (25%) | Improved |
| add feature normalization (ablation) | EliDecide · OOS KCR 25%/50% | F1: 81.46 -> 84.48 (25%); 87.02 -> 88.57 (50%) | Improved |
| add feature normalization (ablation) | EliDecide · OOS KCR 75% | F1: 90.80 vs 90.21 | No clear effect |
| use proposed negative loss with pseudo-open samples instead of L_ADB (ablation) | EliDecide · OOS KCR 25% | F1: 37.42 -> 84.48 | Improved |
| use proposed negative loss instead of L_ADB-Gen (only inside-boundary pseudo-open) (ablation) | EliDecide · OOS KCR 25% | F1: 54.15 -> 84.48 | Improved |
| use ellipsoid EliDecide instead of best baseline | EliDecide · Banking / OOS / StackOverflow avg | F1: 80.91 -> 82.76; 85.22 -> 87.75; 85.78 -> 86.47 | Improved |
| replace BERT backbone with Llama 3 (finetuned) with boundary detection | EliDecide · Banking KCR 25% | F1: 77.75 -> 73.76 | Hurt |
| use ball CF=1 instead of CF=0.975 at low KCR (ablation) | ball boundary · OOS | F1: KCR 25%: 83.20 -> 80.07; KCR 75%: 87.04 -> 89.28 | Mixed |
Closest papers by meaning
- Multi-Granularity Open Intent Classification via Adaptive Granular-Ball Decision Boundary (AAAI 2025 · similarity 0.87)
- MIntRec2.0: A Large-scale Benchmark Dataset for Multimodal Intent Recognition and Out-of-scope Detection in Conversations (ICLR 2024 · similarity 0.85)
- Uni-MIS: United Multiple Intent Spoken Language Understanding via Multi-View Intent-Slot Interaction (AAAI 2024 · similarity 0.82)
- Aligner²: Enhancing Joint Multiple Intent Detection and Slot Filling via Adjustive and Forced Cross-Task Alignment (AAAI 2024 · similarity 0.82)
- Contextual Augmented Global Contrast for Multimodal Intent Recognition (CVPR 2024 · similarity 0.81)
- Concept Matching with Agent for Out-of-Distribution Detection (AAAI 2025 · similarity 0.81)
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.
