Machine learning-based classification of rock discontinuity trace: SMOTE oversampling integrated with GBT ensemble learning

This paper presents a hybrid ensemble classifier combined synthetic minority oversampling technique (SMOTE), random search (RS) hyper-parameters optimization algorithm and gradient boosting tree (GBT) to achieve efficient accurate rock trace identification. A thirteen-dimensional database consisting of basic, vector, discontinuity features is established from image samples. All data points are classified as either “trace” or “non-trace” divide the ultimate results into candidate It found that SMOTE technology can effectively improve classification performance by recommending an optimized imbalance ratio 1:5 1:4. Then, sixteen classifiers generated four basic machine learning (ML) models applied for comparison. The reveal proposed RS-SMOTE-GBT outperforms other fifteen ML algorithms both non-trace classifications. Finally, discussions on feature importance, generalization ability error conducted classifier. experimental indicate more critical affecting primarily features. Besides, cleaning up sedimentary pumice reducing area fractured contribute improving overall performance. method provides new alternative approach identification 3D trace.