A hybrid deep learning-Bayesian optimization model for enhanced slope stability classification
Bibliographic record
Abstract
The process of classifying slopes according to their resistance to failure is important in the geotechnical engineering field for ensuring the safety of infrastructure and human life. This research aims to develop a novel Deep Learning-Bayesian Optimization (DL-BO) model for slope stability classification by determining the best model’s hyperparameters. The study focuses on utilizing a Recurrent Neural Network (RNN), Long Short-Term Memory (LSTM), Bidirectional LSTM (Bi-LSTM), and attention mechanism-enhanced LSTM (Attention-LSTM). In addition, this study applies a Bayesian optimization technique for hyperparameters by using a Gaussian Process (GP) with Expected Improvement (EI) acquisition function to create the non-expensive surrogate model. The data used in the study (Appendix A) consisted of 575 real-life slope samples. The dataset was split into 85:15 training and testing sets, respectively. Five-stratified k-fold cross-validation was used to validate the DL-BO models. The dataset consisted of soil and slope characteristics, including unit weight kN/m 3 , cohesion (kPa), angle of internal friction (degrees), pore water pressure ratio, slope height (m), and slope angle (degrees). Widely used statistical metrics in the research literature were applied to evaluate the DL-BO models, including accuracy, precision, recall, specificity, F1-score , and Area Under the ROC Curve (AUC). RNN-BO and LSTM-BO achieved significant model accuracy, reaching 81.6% and 85.1%, and AUC reaching 89.3% and 89.8%, respectively. While Bi-LSTM-BO enhanced the LSTM model, reaching an accuracy of 87.4% and an AUC of 95.1%, Attention-LSTM-BO arrived at an accuracy of 86.2% and an AUC of 89.6%. The rest of the outcomes are presented in the results and discussion section.
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 | 0.002 |
| Meta-epidemiology (narrow) | 0.001 | 0.001 |
| Meta-epidemiology (broad) | 0.001 | 0.001 |
| Bibliometrics | 0.001 | 0.001 |
| Science and technology studies | 0.000 | 0.000 |
| 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".