Evaluation of a Deep Learning-Based Algorithmic Pipeline for Real-Time Onboard Removal of Glider Self-Noise from Passive Acoustic Data
Bibliographic record
Abstract
Gliders are autonomous underwater vehicles that move by changing their buoyancy and center of gravity, making them a relatively quiet and thus attractive platform for collecting passive acoustic data for many applications, including soundscape/fisheries management, real-time monitoring of endangered whales, and acoustic propagation modeling/validation. Hydrophones can be integrated into the glider for automated analysis and identification from a purpose-built library, but broadband self-noise from internally-moving motors and components must be removed to enhance precise signal processing for targets of interest. Passive acoustic data from two 2024 right whale monitoring missions in shallow coastal waters of the Georgia/Florida calving grounds were used with glider engineering data to develop signal processors and train deep learning algorithms to identify and remove glider self-noise. Additional sampling and data augmentation techniques were integrated to enhance the classification performance of minorities. Based on comparative evaluation, ResNet50 outperformed all other deep learning image classifiers (e.g., EfficientNet, VGG) and was selected as the base deep learning model for selfnoise classification. All models generally performed poorly for all balancing techniques used for classifying the self-noises from the ballast pump, pitch motor, and air pump, likely due to the small minority amount in the baseline training dataset. Results also suggest that oversampling by a large percent increase causes model performance to decrease. The paper highlighted the innovative deep learning framework for onboard glider self-noise removal, which can further reduce false positive detection and enhance whale sound recognition from passive acoustic data.
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 | 0.003 |
| Meta-epidemiology (narrow) | 0.001 | 0.000 |
| Meta-epidemiology (broad) | 0.000 | 0.001 |
| Bibliometrics | 0.001 | 0.000 |
| Science and technology studies | 0.000 | 0.001 |
| Scholarly communication | 0.001 | 0.001 |
| Open science | 0.002 | 0.001 |
| Research integrity | 0.001 | 0.002 |
| Insufficient payload (model declined to judge) | 0.003 | 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".