Visible to the public Biblio

Filters: Keyword is domain experts  [Clear All Filters]
2021-03-01
Kuppa, A., Le-Khac, N.-A..  2020.  Black Box Attacks on Explainable Artificial Intelligence(XAI) methods in Cyber Security. 2020 International Joint Conference on Neural Networks (IJCNN). :1–8.

Cybersecurity community is slowly leveraging Machine Learning (ML) to combat ever evolving threats. One of the biggest drivers for successful adoption of these models is how well domain experts and users are able to understand and trust their functionality. As these black-box models are being employed to make important predictions, the demand for transparency and explainability is increasing from the stakeholders.Explanations supporting the output of ML models are crucial in cyber security, where experts require far more information from the model than a simple binary output for their analysis. Recent approaches in the literature have focused on three different areas: (a) creating and improving explainability methods which help users better understand the internal workings of ML models and their outputs; (b) attacks on interpreters in white box setting; (c) defining the exact properties and metrics of the explanations generated by models. However, they have not covered, the security properties and threat models relevant to cybersecurity domain, and attacks on explainable models in black box settings.In this paper, we bridge this gap by proposing a taxonomy for Explainable Artificial Intelligence (XAI) methods, covering various security properties and threat models relevant to cyber security domain. We design a novel black box attack for analyzing the consistency, correctness and confidence security properties of gradient based XAI methods. We validate our proposed system on 3 security-relevant data-sets and models, and demonstrate that the method achieves attacker's goal of misleading both the classifier and explanation report and, only explainability method without affecting the classifier output. Our evaluation of the proposed approach shows promising results and can help in designing secure and robust XAI methods.

2020-10-12
Eckhart, Matthias, Ekelhart, Andreas, Lüder, Arndt, Biffl, Stefan, Weippl, Edgar.  2019.  Security Development Lifecycle for Cyber-Physical Production Systems. IECON 2019 - 45th Annual Conference of the IEEE Industrial Electronics Society. 1:3004–3011.

As the connectivity within manufacturing processes increases in light of Industry 4.0, information security becomes a pressing issue for product suppliers, systems integrators, and asset owners. Reaching new heights in digitizing the manufacturing industry also provides more targets for cyber attacks, hence, cyber-physical production systems (CPPSs) must be adequately secured to prevent malicious acts. To achieve a sufficient level of security, proper defense mechanisms must be integrated already early on in the systems' lifecycle and not just eventually in the operation phase. Although standardization efforts exist with the objective of guiding involved stakeholders toward the establishment of a holistic industrial security concept (e.g., IEC 62443), a dedicated security development lifecycle for systems integrators is missing. This represents a major challenge for engineers who lack sufficient information security knowledge, as they may not be able to identify security-related activities that can be performed along the production systems engineering (PSE) process. In this paper, we propose a novel methodology named Security Development Lifecycle for Cyber-Physical Production Systems (SDL-CPPS) that aims to foster security by design for CPPSs, i.e., the engineering of smart production systems with security in mind. More specifically, we derive security-related activities based on (i) security standards and guidelines, and (ii) relevant literature, leading to a security-improved PSE process that can be implemented by systems integrators. Furthermore, this paper informs domain experts on how they can conduct these security-enhancing activities and provides pointers to relevant works that may fill the potential knowledge gap. Finally, we review the proposed approach by means of discussions in a workshop setting with technical managers of an Austrian-based systems integrator to identify barriers to adopting the SDL-CPPS.

2018-09-05
Teusner, R., Matthies, C., Giese, P..  2017.  Should I Bug You? Identifying Domain Experts in Software Projects Using Code Complexity Metrics 2017 IEEE International Conference on Software Quality, Reliability and Security (QRS). :418–425.
In any sufficiently complex software system there are experts, having a deeper understanding of parts of the system than others. However, it is not always clear who these experts are and which particular parts of the system they can provide help with. We propose a framework to elicit the expertise of developers and recommend experts by analyzing complexity measures over time. Furthermore, teams can detect those parts of the software for which currently no, or only few experts exist and take preventive actions to keep the collective code knowledge and ownership high. We employed the developed approach at a medium-sized company. The results were evaluated with a survey, comparing the perceived and the computed expertise of developers. We show that aggregated code metrics can be used to identify experts for different software components. The identified experts were rated as acceptable candidates by developers in over 90% of all cases.
2018-03-19
Thankaraj, A., Nair, A. J., Vasudevan, N., Pathari, V..  2017.  Misclassifications: The Missing Link. 2017 International Conference on Advances in Computing, Communications and Informatics (ICACCI). :1719–1722.

The notion of style is pivotal to literature. The choice of a certain writing style moulds and enhances the overall character of a book. Stylometry uses statistical methods to analyze literary style. This work aims to build a recommendation system based on the similarity in stylometric cues of various authors. The problem at hand is in close proximity to the author attribution problem. It follows a supervised approach with an initial corpus of books labelled with their respective authors as training set and generate recommendations based on the misclassified books. Results in book similarity are substantiated by domain experts.