Uncertainty Quantification of Different Data Sources with Regard to a LSTM Analysis of Grinded Surfaces
Bibliographic record
Abstract
To improve the conventional methods of condition monitoring, a new image processing analysis approach is needed to get a faster and more cost-effective analysis of produced surfaces.For this reason, different optical techniques based on image analysis have been developed over the past years.In this study, fine grinded surface images have been generated under constant boundary conditions in a test rig built up in a lab.The gathered image material in combination with the classical measured surface topography values is used as the training data for machine learning analyses.The image of each grinded surface is analyzed regarding its measured arithmetic average roughness value (Ra) by the use of Recurrent Neural Networks (in this case LSTM).LSTMs are a type of machine learning algorithms which can particularly be applied for any kind of analysis based on time series.In this paper a possible optimization potential of the available databases is analyzed.For this purpose, two different sets of images with various resolutions were taken under the same conditions.Since the data plays an essential role for the training of machine learning models, the challenge in the application is often to find costefficient, fast and at the same time process-adaptable measurement methods that also have sufficient accuracy.Thus, the target values recorded with tactile measurement method are compared to a more precise confocal / optical measurement method.This results in two data sets with unequal distributions and different statistical variance.
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 | 0.007 |
| Meta-epidemiology (narrow) | 0.001 | 0.000 |
| Meta-epidemiology (broad) | 0.001 | 0.001 |
| Bibliometrics | 0.001 | 0.001 |
| 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".