English (United Kingdom)

https://curated-unify.zendy.io/wp-json/zendy-region/v1/featured_content/oa?rat=en

https://curated-unify.zendy.io/wp-json/zendy-region/v1/highlighted_journal/

Global: Zendy Plus annual

Presents the access of premium content as premium feature

Premium Content

Presents the keyphrase highlighting as premium feature

Keyphrase Highlighting

Presents the summarisation as premium feature

Summarisation

Insights

Presents the pdf analysis as premium feature

PDF Analysis

Presents the zaia usage as premium feature

ZAIA

Global: Zendy Plus monthly

Global: Zendy Tools annual

Global: Zendy Tools monthly

Global: Zendy Free Plan

Interpretability, trustworthiness, and usability are key considerations inhigh-stake security applications, especially when utilizing deep learningmodels. While these models are known for their high accuracy, they behave asblack boxes in which identifying important features and factors that led to aclassification or a prediction is difficult. This can lead to uncertainty anddistrust, especially when an incorrect prediction results in severeconsequences. Thus, explanation methods aim to provide insights into the innerworking of deep learning models. However, most explanation methods provideinconsistent explanations, have low fidelity, and are susceptible toadversarial manipulation, which can reduce model trustworthiness. This paperprovides a comprehensive analysis of explainable methods and demonstrates theirefficacy in three distinct security applications: anomaly detection usingsystem logs, malware prediction, and detection of adversarial images. Ourquantitative and qualitative analysis reveals serious limitations and concernsin state-of-the-art explanation methods in all three applications. We show thatexplanation methods for security applications necessitate distinctcharacteristics, such as stability, fidelity, robustness, and usability, amongothers, which we outline as the prerequisites for trustworthy explanationmethods.

SoK: Modeling Explainability in Security Analytics for Interpretability, Trustworthiness, and Usability