Xiaofei Xie

TL;DR: Wang et al. as mentioned in this paper proposed principled guidelines to assess the quality of five explanation approaches by designing three critical quantitative metrics to measure their stability, robustness, and effectiveness, and collected five widely-used malware datasets and applied the explanation approaches on them in two tasks, including malware detection and familial identification.

TL;DR: The authors proposed an adversarial attack based on metamorphic relations on the Turing-complete differentiable neural computer (DNC) to expose weaknesses and susceptibilities in modern neural reasoning architectures.

TL;DR: In this paper, a fine-grained program analysis is used to identify available variables and infer types of these variables for the mutation, and an automatic repair strategy is proposed to repair syntax/semantic errors in invalid test cases.

TL;DR: A quantitative analysis framework - DeepStellar - to pave the way for effective quality and security analysis of software systems powered by RNNs, which outperforms existing approaches three hundred times in generating defect-triggering tests and achieves 97% accuracy in detecting adversarial attacks.

TL;DR: DeepSonar as discussed by the authors leverages the power of layerwise neuron activation patterns with a conjecture that they can capture the subtle differences between real and AI-synthesized fake voices, in providing a cleaner signal to classifiers than raw inputs.