Research map / Label-efficient and robust learning / Semi-supervised learning
OpenMix: Reviving Known Knowledge for Discovering Novel Visual Categories in an Open World
Check 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 |
|---|---|---|---|
| add OpenMix to the baseline clustering model | novel class discovery model · CIFAR-10 | ACC: 90.9% -> 95.3% | Helped |
| add OpenMix to the baseline clustering model | novel class discovery model · CIFAR-100 | ACC: 81.2% -> 87.2% | Helped |
| add OpenMix to the baseline clustering model | novel class discovery model · ImageNet | ACC: 77.1% -> 85.7% | Helped |
| remove pseudo-pair learning (PPL) (ablation) | Baseline · CIFAR-10 | ACC: 90.9% -> 70.8% | Hurt |
| remove pseudo-pair learning (PPL) (ablation) | Baseline · CIFAR-100 | ACC: 81.2% -> 23.9% | Hurt |
| remove pseudo-label learning (PLL) (ablation) | Baseline · CIFAR-100 | ACC: 81.2% -> 77.2% | Hurt |
| apply original MixUp between unlabeled samples (ablation) | Baseline · CIFAR-10 | ACC: 90.9% -> 80.2% | Hurt |
| apply original MixUp between unlabeled samples (ablation) | Baseline · CIFAR-100 | ACC: 81.2% -> 78.2% | Hurt |
| extend labeled and unlabeled label distributions to joint classes (ablation) | Baseline · CIFAR-100 | ACC: 81.2% -> 81.9% | Helped |
| add OpenMix mixing labeled examples with unlabeled examples (ablation) | Baseline · CIFAR-10 | ACC: 90.9% -> 93.3% | Helped |
| add OpenMix mixing labeled examples with unlabeled examples (ablation) | Baseline · CIFAR-100 | ACC: 81.2% -> 84.5% | Helped |
| add OpenMix mixing reliable anchors with unlabeled samples (ablation) | OpenMix w/o A · CIFAR-10 | ACC: 93.3% -> 95.3% | Helped |
Closest papers by meaning
- Semantic-Guided Novel Category Discovery (AAAI 2024 · similarity 0.88)
- ProxyDet: Synthesizing Proxy Novel Classes via Classwise Mixup for Open-Vocabulary Object Detection (AAAI 2024 · similarity 0.85)
- Fine-grained Classes and How to Find Them (ICML 2024 · similarity 0.85)
- Envisioning Class Entity Reasoning by Large Language Models for Few-shot Learning (AAAI 2025 · similarity 0.85)
- Novel Class Discovery for Ultra-Fine-Grained Visual Categorization (CVPR 2024 · similarity 0.85)
- Navigating Real-World Partial Label Learning: Unveiling Fine-Grained Images with Attributes (AAAI 2024 · similarity 0.85)
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.