Research map / Label-efficient and robust learning / Semi-supervised learning
A Semi-Supervised Deep Rule-Based Approach for Complex Satellite Sensor Image Analysis
Xiaowei Gu, Plamen Parvanov Angelov, Ce Zhang, Peter M. Atkinson
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 chunk-wise semi-supervised pseudo-labelling over DRB baseline | SeRBIA · WHU-RS 10% labelled | accuracy: 0.8454 -> 0.9006 | Improved |
| add chunk-wise semi-supervised learning over DRB baseline | SeRBIA · NWPU45 10% labelled | accuracy: 0.7411 -> 0.8732 | Improved |
| add chunk-wise semi-supervised learning over DRB baseline | SeRBIA · AID 20% labelled (SOTA protocol) | accuracy: 0.8358 -> 0.9429 | Improved |
| add chunk-wise semi-supervised learning over DRB baseline | SeRBIA · UCMerced 10% labelled | accuracy: 0.8025 -> 0.8871 | Improved |
| use SeRBIA instead of LGC semi-supervised baseline | SeRBIA · RSSCN7 10% labelled | accuracy: 0.8848 -> 0.8740 | Hurt |
| use SeRBIA instead of LGC | SeRBIA · AID 10% labelled | accuracy: 0.8290 -> 0.8053 | Hurt |
| use SeRBIA instead of TSDA-ELM | SeRBIA · WHU-RS 60% labelled | accuracy: 0.9892 -> 0.9802 | Hurt |
| increase labelled ratio from 10% to 20% | SeRBIA · AID | accuracy: 0.8053 -> 0.8351 | Improved |
| lower phi (looser pseudo-label rigour) (ablation) | SeRBIA · Singapore/UCMerced/WHU-RS/RSSCN7 (10% labelled) | accuracy | Hurt |
| raise phi (stricter pseudo-labelling) (ablation) | SeRBIA · Singapore/UCMerced/WHU-RS/RSSCN7 (10% labelled) | accuracy | Hurt |
| increase gamma (more sensitivity to new patterns) (ablation) | SeRBIA · Singapore/UCMerced/WHU-RS/RSSCN7 (10% labelled) | accuracy | Improved |
| vary chunk size W from 100 to 800 (ablation) | SeRBIA · Singapore/UCMerced/WHU-RS/RSSCN7 (10% labelled) | accuracy | No clear effect |
Closest papers by meaning
- SkySense: A Multi-Modal Remote Sensing Foundation Model Towards Universal Interpretation for Earth Observation Imagery (CVPR 2024 · similarity 0.85)
- Galileo: Learning Global & Local Features of Many Remote Sensing Modalities (ICML 2025 · similarity 0.84)
- Parameter Efficient Self-Supervised Geospatial Domain Adaptation (CVPR 2024 · similarity 0.84)
- Learning De-Biased Representations for Remote-Sensing Imagery (NeurIPS 2024 · similarity 0.84)
- InstructSAM: A Training-free Framework for Instruction-Oriented Remote Sensing Object Recognition (NeurIPS 2025 · similarity 0.84)
- Rethinking Transformers Pre-training for Multi-Spectral Satellite Imagery (CVPR 2024 · 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.
