Search for: All records

Abstract Unlike micromechanics failure models that have a well-defined crack path, phase-field fracture models are capable of predicting the crack path in arbitrary geometries and dimensions by utilizing a diffuse representation of cracks. However, such models rely on the calibration of a fracture energy (Gc) and a regularization length-scale (lc) parameter, which do not have a strong micromechanical basis. Here, we construct the equivalent crack-tip cohesive zone laws representing a phase-field fracture model, to elucidate the effects of Gc and lc on the fracture resistance and crack growth mechanics under mode I K-field loading. Our results show that the cohesive zone law scales with increasing Gc while maintaining the same functional form. In contrast, increasing lc broadens the process zone and results in a flattened traction-separation profile with a decreased but sustained peak cohesive traction over longer separation distances. While Gc quantitatively captures the fracture initiation toughness, increasing Gc coupled with decreasing lc contributes to a rising fracture resistance curve and a higher steady-state toughness—both these effects cumulate in an evolving cohesive zone law with crack progression. We discuss the relationship between these phase-field parameters and process zone characteristics in the material. 

This paper introduces robustness verification for semantic segmentation neural networks (in short, semantic segmentation networks [SSNs]), building on and extending recent approaches for robustness verification of image classification neural networks. Despite recent progress in developing verification methods for specifications such as local adversarial robustness in deep neural networks (DNNs) in terms of scalability, precision, and applicability to different network architectures, layers, and activation functions, robustness verification of semantic segmentation has not yet been considered. We address this limitation by developing and applying new robustness analysis methods for several segmentation neural network architectures, specifically by addressing reachability analysis of up-sampling layers, such as transposed convolution and dilated convolution. We consider several definitions of robustness for segmentation, such as the percentage of pixels in the output that can be proven robust under different adversarial perturbations, and a robust variant of intersection-over-union (IoU), the typical performance evaluation measure for segmentation tasks. Our approach is based on a new relaxed reachability method, allowing users to select the percentage of a number of linear programming problems (LPs) to solve when constructing the reachable set, through a relaxation factor percentage. The approach is implemented within NNV, then applied and evaluated on segmentation datasets, such as a multi-digit variant of MNIST known as M2NIST. Thorough experiments show that by using transposed convolution for up-sampling and average-pooling for down-sampling, combined with minimizing the number of ReLU layers in the SSNs, we can obtain SSNs with not only high accuracy (IoU), but also that are more robust to adversarial attacks and amenable to verification. Additionally, using our new relaxed reachability method, we can significantly reduce the verification time for neural networks whose ReLU layers dominate the total analysis time, even in classification tasks.