Classification with U-Net neural networks of herring, salmon, and bubbles in multifrequency echograms
Bibliographic record
Abstract
Convolutional neural networks were used to accurately identify biological and physical phenomena in multifrequency echograms. The primary classes of interest for this study were herring and juvenile salmon. Secondary classes included gas bubbles and the air-sea interface. A collection of U-Nets were trained to apply pixel-level classification. In addition to recorded acoustic data at four frequencies (67.5, 125, 200, 455 kHz), simulated data were created to provide context for the interpretation of the echograms. This contextual data included a channel for water depth and another channel for solar elevation angle. It is demonstrated that the inclusion of these context channels improve the efficacy of the classifier. After the pixel-level classification was performed, traditional school detection methods were applied to classify distinct schools of herring and aggregations of salmon. The pixel-level classification results were refined using the mean classification score of each school. To ensure broad applicability, this network was trained to classify echograms with noise left intact. It is also demonstrated that this method is scalable and effective at classifying data with a lower sampling rate than that used for training. The best performing model classified herring, salmon, and bubble classes with F1 scores of 93.0%, 87.3%, and 86.5%, respectively.
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 | 0.001 | 0.002 |
| Meta-epidemiology (narrow) | 0.001 | 0.000 |
| Meta-epidemiology (broad) | 0.000 | 0.000 |
| Bibliometrics | 0.001 | 0.000 |
| Science and technology studies | 0.000 | 0.000 |
| 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 | 0.000 |
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 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".