Machine learning in concrete durability: challenges and pathways identified by RILEM TC 315-DCS towards enhanced predictive models
Bibliographic record
Abstract
Abstract This review provides an in-depth examination of machine learning applications in assessing concrete durability from 2013 to 2024, with a particular focus on critical degradation mechanisms, including carbonation, chloride-induced deterioration, sulfate attack, frost damage, shrinkage, and corrosion. It underscores the field’s heavy reliance on laboratory-based data and notes the limited use of field data and the scarcity of newly generated datasets. The review reveals that most studies utilize existing literature-based datasets, with few contributing novel data and limited open access to these databases, which hampers broader validation and application. The review classifies the features analyzed in studies into categories such as mixture proportions, engineering properties, exposure conditions, test parameters, and chemical compositions, highlighting a growing emphasis on chemical compositions. Modeling approaches are predominantly standalone, though ensemble and hybrid models are increasingly prevalent, with ensemble models showing particularly strong performance in recent years. High accuracy is observed across studies, with ensemble models, neural networks, and hybrid models leading in performance. Furthermore, the review stresses the growing importance of model explainability, noting that model-agnostic methods like SHAP are frequently used and that the focus on explainability has increased. To propel the field forward, the review advocates for the development of diverse new datasets that include both the chemical and physical properties of various mix ingredients and improved data-sharing practices. It recommends adopting a multi-task learning approach to simultaneously address multiple deterioration mechanisms, which can yield deeper insights and support the creation of more durable concrete structures.
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 | 0.006 | 0.011 |
| Meta-epidemiology (narrow) | 0.001 | 0.000 |
| Meta-epidemiology (broad) | 0.001 | 0.001 |
| Bibliometrics | 0.001 | 0.002 |
| Science and technology studies | 0.000 | 0.001 |
| Scholarly communication | 0.004 | 0.003 |
| Open science | 0.001 | 0.002 |
| Research integrity | 0.001 | 0.003 |
| Insufficient payload (model declined to judge) | 0.005 | 0.002 |
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".