Effect of Dataset Size and Auxiliary Data in Bayesian Learning of Advanced Manufacturing: A Composite Autoclave Processing Diagnostic Study
Bibliographic record
Abstract
Abstract Recent advances in data-driven predictive modelling have enabled the emergence of intelligent manufacturing enterprises. Nonetheless, most of the present frameworks incorporate non-interpretable black-box machine learning models, often requiring large datasets and yet lacking ‘diagnostic’ modelling capabilities. In the context of advanced composites manufacturing, where the presence of numerous decision factors and uncertainties can rapidly yield failures, training cost/data-efficient, transparent and diagnostic-capable predictive models continue to highly valuable to pertinent industries. This can specifically allow decision-makers on the manufacturing floor to identify the causes or state variables of the process that contribute to the product failure (e.g., due to an excessive exotherm or lag temperature during curing), and thereby saving sizable volume of material scraps due to trial and errors. In this work, a Bayesian learning framework with inverse modelling capabilities for an advanced composites autoclave curing process has been developed and assessed for the first time, while assuming different dataset size availabilities. The advantages of using both a naïve Bayesian network and a highly-connected Bayesian belief network (BBN) are compared and discussed. The results revealed that integration of expert knowledge under highly-connected Bayesian models can offer a favorable predictive performance for root cause analyses, along with apparent tractability for in-situ applications, despite the very limited-volume of training data, when accompanied with carefully selected auxiliary data (e.g. via the use of a proxy thermocouple during the processing based on expert domain).
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 | 0.011 | 0.035 |
| Meta-epidemiology (narrow) | 0.001 | 0.000 |
| Meta-epidemiology (broad) | 0.001 | 0.001 |
| Bibliometrics | 0.001 | 0.001 |
| Science and technology studies | 0.001 | 0.001 |
| Scholarly communication | 0.001 | 0.002 |
| Open science | 0.001 | 0.001 |
| Research integrity | 0.002 | 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".