Accuracy metrics have been widely used for the evaluation of predictions in machine learning. However, the selection of an appropriate accuracy metric for the evaluation of a specific prediction has not yet been specified. In this study, seven of the most used accuracy metrics in machine learning were summarized, and both their advantages and disadvantages were studied. To achieve this, the acoustic emission data of damage locations were collected from a pile hit test. A backpropagation artificial neural network prediction model for damage locations was trained with acoustic emission data using six different training algorithms, and the prediction accuracies of six algorithms were evaluated using seven different accuracy metrics. Test results showed that the training algorithm of “TRAINGLM” exhibited the best performance for predicting damage locations in deep piles. Subsequently, the artificial neural networks were trained using three different datasets collected from three acoustic emission sensor groups, and the prediction accuracies of three models were evaluated with the seven different accuracy metrics. The test results showed that the dataset collected from the pile body-installed sensors group exhibited the highest accuracy for predicting damage locations in deep piles. Subsequently, the correlations between the seven accuracy metrics and the sensitivity of each accuracy metrics were discussed based on the analysis results. Eventually, a novel selection method for an appropriate accuracy metric to evaluate the accuracy of specific predictions was proposed. This novel method is useful to select an appropriate accuracy metric for wide predictions, especially in the engineering field.

Understanding the shear behavior of the interface between surrounding rock and backfill is of significant engineering importance for enhancing stope stability in cemented tailings backfill mining. However, the evolutionary mechanisms of shear properties and damage under varying mechanical conditions remain insufficiently studied. This investigation employed tailings and surrounding rock from a Guangdong tailings pond, with basic mechanical parameters determined through laboratory tests. Numerical models of the rock-backfill composite were developed using PFC2D, considering different shear rates (0.3, 0.6, and 0.9 mm/min), lateral confinement levels (0.5, 1.0, and 1.5 MPa), and roughness coefficients. The analysis compared the interface’s peak and residual shear strengths, revealed crack evolution patterns, and explored damage mechanisms using acoustic emission monitoring and energy dissipation theory. Key findings include the following: (1) Shear stress–displacement curves under all conditions exhibited three stages, ascending, shearing-off, and sliding, with distinct peak and residual strengths. (2) Increasing lateral confinement, shear rate, and roughness transformed failure from localized to global sliding, with cracks occurring at the interface and propagating into the backfill. (3) Cumulative acoustic emission events increased with all three factors, with lateral confinement showing the most substantial effect on interface energy accumulation (83% increase). These results provide theoretical support for assessing interface stability in deep backfilled stopes.