Journal Article10.1109/ecrime61234.2023.10485502
Achieving High-Fidelity Explanations for Risk Exposition Assessment in the Cybersecurity Domain
Albert Grau Calvo,Santiago Escuder,J. Escrig,Xavier Marrugat,Nil Ortiz,Jordi Guijarro +5 more
- 15 Nov 2023
pp 1-10
TL;DR: The proposed framework offers a swift and dependable method for assessing explanations specifically tailored for the cybersecurity domain and proposes an explainable proxy which is founded on the generation of systematic evaluations of explanations.
read more
Abstract: Understanding AI-driven systems has become fundamental, particularly when these systems are employed for critical decision-making, as is the case in the field of cybersecurity. In this regard, explainability has been extensively advocated as a cornerstone to comprehend the model, thereby enhancing trust and accountability in data-driven systems. Through the successful use-case of a risk exposure assessment framework which aims to proactively reduce an organization's attack surface, we propose an explainable proxy which is founded on the generation of systematic evaluations of explanations. The proposed framework offers a swift and dependable method for assessing explanations specifically tailored for the cybersecurity domain.
read more
Chat with Paper
AI Agents for this Paper
Find similar papers on Google Scholar, PubMed and Arxiv
Write a critical review of this paper
Analyze citations of this paper to find unaddressed research gaps
References
An Explainable AI-Based Intrusion Detection System for DNS Over HTTPS (DoH) Attacks
TL;DR: This paper has used the publicly available CIRA-CIC-DoHBrw-2020 dataset for developing an accurate solution to detect and classify the DNS over HTTPS attacks and proposed balanced and stacked Random Forest achieved very high precision, recall, and F1 score for the classification task at hand.
47
Robust Network Intrusion Detection Through Explainable Artificial Intelligence (XAI)
01 Sep 2022
TL;DR: In this paper , an extreme gradient boosting (XGBoost) model is used to perform intrusion detection and an auto-encoder is trained to distinguish previously seen and unseen attacks.
45
OpenXAI: Towards a Transparent Evaluation of Model Explanations
22 Jun 2022
TL;DR: OpenXAI as mentioned in this paper is an open source framework for evaluating and benchmarking post-hoc explanation methods, which includes a flexible synthetic data generator and a collection of diverse real-world datasets and pre-trained models.
Detecting vulnerabilities in IoT software: New hybrid model and comprehensive data analysis
TL;DR: Wang et al. as discussed by the authors proposed a contextual embedding model to integrate three hybrid models, CLSTM, CBiLSTM (sequential structure), and CNN-BiLSTMs (parallel structure), based on the code characteristics of IoT applications.
11
•Posted Content
On the Importance of Domain-specific Explanations in AI-based Cybersecurity Systems (Technical Report).
TL;DR: In this paper, the authors make three contributions: (i) proposal and discussion of desiderata for the explanation of outputs generated by AI-based cybersecurity systems; (ii) a comparative analysis of approaches in the literature on explainable artificial intelligence (XAI) under the lens of both our desidersata and further dimensions that are typically used for examining XAI approaches; and (iii) a general architecture that can serve as a roadmap for guiding research efforts towards the development of explainable AI-Based cybersecurity systems.
6