From Dual to Qual: A Feature-Analysis-Oriented Interpretable Polarization Feature Generative Mapping Model for SAR Oil Spill Detection
Why this work is in the frame
A frame that forgets how it found something cannot be audited. These are the routes that admitted this work.
Bibliographic record
Abstract
Oil spills can cause serious pollution to the marine environment. Synthetic aperture radar (SAR), as an all-day-all-weather active microwave sensor, can provide a powerful solution for oil spill detection. However, due to the limitations of system characteristics, data-information imbalance problem exists in research based on polarimetric SAR. To address the above problems, a polarization feature generative mapping model (PF-GMM) for oil spill detection tasks is proposed in this paper. PF-GMM maps dual-polarization features (DPFs) to qual-polarization features (QPFs) through a generative adversarial approach. To select DPFs that can cover qual-polarization information, an Interpretable Analysis Module (IAM) was designed. IAM analyzed the feature contribution and the interaction between DPFs and QPFs to reveal the significance of each DPF in model optimization and physical level, so as to achieve the optimal selection of DPFs. Based on the selected domain feature group (DFG), a Dual-Pol-SAR Oil Spill Dataset (DPSOS) was constructed to evaluate the performance of the selected features and make up for the lack of SAR oil spill detection dataset. Experimental results show that DFG can effectively achieve oil spill segmentation in different scenarios, and to a certain extent achieves oil spill detection performance similar to that of QPFs. PF-GMM simultaneously ensures the superiority of the selected features in engineering applications and physical meanings. Its results can cover qual-polarization information to the greatest extent, make up for the defects and deficiencies of dual-polarization data, and have strong engineering guidance value.
Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.
Full frame distilled prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. Learned from the 10,348 direct Codex labels and 10,348 direct Gemma labels. Candidate is the union of thresholded teacher heads; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels or direct frontier model labels.
Codex and Gemma teacher scores by category
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.000 | 0.000 |
| Meta-epidemiology (narrow) | 0.001 | 0.001 |
| Meta-epidemiology (broad) | 0.001 | 0.001 |
| Bibliometrics | 0.001 | 0.004 |
| Science and technology studies | 0.001 | 0.000 |
| Scholarly communication | 0.000 | 0.001 |
| Open science | 0.000 | 0.000 |
| Research integrity | 0.001 | 0.001 |
| Insufficient payload (model declined to judge) | 0.000 | 0.000 |
Machine scores (provisional)
The two teacher heads of the student model, read on this work. A score orders the frame for review; it never asserts a category, and the validation status ships verbatim with every row.
Baseline scores from an immature model (maturity gate not passed, 7 training rounds). Scores rank; they never assert a category.
score_only:v0-immature-baseline · verbatim from the scoring run: score_only means the number may rank works, and no category label ships from it