Research map / Label-efficient and robust learning / Recognition and long-tail
Learning Structured Inference Neural Networks With Label Relations
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 |
|---|---|---|---|
| Use structured positive and negative label relations rather than an unstructured BINN | CNN label prediction on AwA · AwA, 50 animal classes | IoU Acc: 62.80±0.25 -> 66.60±0.43 | Helped |
| Use structured label relations rather than logistic classification alone | CNN label prediction on AwA · AwA, 28 taxonomy terms | mAP_L: 90.16±0.10 -> 93.00±0.29 | Helped |
| Add group-label observations to the tag-observed SINN (ablation) | SINN on NUS-WIDE · NUS-WIDE | mAP_L: 67.20±0.60 -> 69.24±0.47 | Helped |
| Use SINN with group and tag observations rather than logistic regression with the same metadata | NUS-WIDE classification · NUS-WIDE | mAP_L: 54.67±0.57 -> 69.24±0.47 | Helped |
| Add structured label relations rather than logistic classification alone | CNN label prediction on SUN397 · SUN397, 397 fine-grained scene categories | mAP_L: 55.31±0.30 -> 58.00±0.33 | Helped |
| Add observed coarse scene labels to SINN (ablation) | CNN + SINN on SUN397 · SUN397, 397 fine-grained scene categories | MC Acc: 57.60±0.38 -> 63.46±0.18 | Helped |
| Add observed coarse scene labels to logistic classification (ablation) | CNN + Logistics on SUN397 · SUN397, 397 fine-grained scene categories | MC Acc: 57.86±0.38 -> 59.08±0.27 | Helped |
| Fine-tune the first seven CNN layers with SINN (ablation) | CNN + SINN on AwA · AwA, 28 taxonomy terms | IoU Acc: 84.47±0.38 -> 86.06±0.72 | Helped |
Closest papers by meaning
- Multi-Label Zero-Shot Learning With Structured Knowledge Graphs (CVPR 2018 · similarity 0.87)
- CNN-RNN: A Unified Framework for Multi-Label Image Classification (CVPR 2016 · similarity 0.87)
- Multi-Label Image Recognition With Graph Convolutional Networks (CVPR 2019 · similarity 0.87)
- Structured Prediction Energy Networks (ICML 2016 · similarity 0.86)
- Learning Spatial Regularization With Image-Level Supervisions for Multi-Label Image Classification (CVPR 2017 · similarity 0.86)
- MLC-NC: Long-Tailed Multi-Label Image Classification Through the Lens of Neural Collapse (AAAI 2025 · similarity 0.84)
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.