Effect of training sample size, image resolution and epochs on filamentous and floc-forming bacteria classification using machine learning
Bibliographic record
Abstract
Computer vision techniques can expedite the detection of bacterial growth in wastewater treatment plants and alleviate some of the shortcomings associated with traditional detection methods. In recent years, researchers capitalized on this potential by developing segmentation algorithms that were specifically tailored to identify the overgrowth of filamentous bacteria and the risk of sludge bulking. This study investigated the optimization of an artificial intelligence (AI) segmentation model in terms of accuracy metrics and computational requirements. Specifically, three model variables were tested, including training sample size, image resolution, and number of training epochs. The results indicated that larger sample sizes resulted in higher output accuracy up to a certain limit (300 images), beyond which no significant improvements were observed. High image resolution (788 × 530) provided more details for the deep learning model to detect the fine edges between bacteria albeit with significant additional computational requirements. The addition of more training epochs resulted in a minor increase in segmentation accuracy, particularly for thin interconnected filamentous bacteria. Overall, high resolution and epochs did not have a major effect when the sample size was large (300 and 500 images). The findings highlight the optimal balance between model accuracy and computational demands, emphasizing the importance of prioritizing diverse training samples with sufficient sample size. This approach is critical for large-scale implementation, as it enhances the potential of AI to deliver timely and accurate predictions, leading to early warnings of wastewater treatment issues such as sludge bulking. • A deep learning image classification model was trained and optimized on image data. • Three parameters were tested: sample size, image resolution, and training epochs. • Larger sample sizes improved accuracy up to 300 images, beyond which gains plateaued. • High image resolution improved edge detection with significant computation requirements. • More training epochs slightly enhanced accuracy, especially for filamentous bacteria.
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 | 0.003 | 0.012 |
| Meta-epidemiology (narrow) | 0.001 | 0.000 |
| Meta-epidemiology (broad) | 0.001 | 0.001 |
| Bibliometrics | 0.000 | 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 | 0.000 |
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".