Diffraction denoising using self‐supervised learning
Bibliographic record
Abstract
Abstract Diffraction wavefield contains valuable information on subsurface composition through velocity extraction and sometimes anisotropy estimation. It can also be used for the delineation of geological features, such as faults, fractures and mineral deposits. Diffraction recognition is, therefore, crucial for improved interpretation of seismic data. To date, many workflows for diffraction denoising, including deep‐learning applications, have been provided, however, with a major focus on sedimentary settings or for ground‐penetrating radar data. In this study, we have developed a workflow for a self‐supervised learning technique, an autoencoder, for diffraction denoising on synthetic seismic, ground‐penetrating radar and hardrock seismic datasets. The autoencoder provides promising results especially for the ground‐penetrating radar data. Depending on the target of the studies, diffraction signals can be tackled using the autoencoder both as the signal and/or noise when, for example, a reflection is a target. The real hardrock seismic data required additional pre‐ and post‐autoencoder image processing steps to improve automatic delineation of the diffraction. Here, we also coupled the autoencoder with Hough transform and pixel edge detection filters. Along inlines and crosslines, diffraction signals have sometimes a similar character as the reflection and may spatially be correlated making the denoising workflow unsuccessful. Coupled with additional image processing steps, we successfully isolated diffraction that is generated from a known volcanogenic massive sulphide deposit. These encouraging results suggest that the self‐supervised learning techniques such as the autoencoder can be used also for seismic mineral exploration purposes and are worthy to be implemented as additional tools for data processing and target detections.
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.
How this classification was reachedexpand
Full frame machine prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. The Gemma side is a direct model label for every work in the frame, read from the title-only record. The Codex side is a classifier learned from the 10,348 direct Codex labels and calibrated to design-weighted sample rates; fields without enough sample support carry no Codex call. Candidate is the union of the two sides; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels.
Distilled classifier scores by category (both heads)
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.002 | 0.004 |
| Meta-epidemiology (narrow) | 0.001 | 0.000 |
| Meta-epidemiology (broad) | 0.001 | 0.001 |
| Bibliometrics | 0.001 | 0.000 |
| Science and technology studies | 0.000 | 0.001 |
| Scholarly communication | 0.001 | 0.001 |
| Open science | 0.001 | 0.001 |
| Research integrity | 0.001 | 0.001 |
| Insufficient payload (model declined to judge) | 0.001 | 0.001 |
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 itClassification
machine, unvalidatedMachine predicted; a candidate call from one source (direct Gemma or distilled Codex), not a consensus.
How this classification was reached, model by model and score by score, is at the end of the page under "How this classification was reached".