Synapse
⌘+K
Synapse
PulseExploreClubsResearchersJournals
Instagram
HomeClubsExplore
April 12, 2026Information Systems FrontiersOpen Access

Toward Human-Centered Explainability: Natural Language Explanations for Anomaly Detection

View Full Paper
Ask AI
Bookmark
Share

Authors

HPHéctor Padín-TorrenteVCVictor Carneiro-DiazIOInes Ortega-Fernandez

Discussion

Loading...

Member takes

Overview

This work proposes a human-centered AI pipeline to improve anomaly detection explanations for cybersecurity, indicating better alignment with expert decision-making.

Key Points

  • The aim is to develop an explainable AI pipeline that provides understandable explanations for anomaly detection in cybersecurity contexts.
  • Developed a human-centered explainable AI pipeline for anomaly detection.
  • Utilized local large language models to generate contextual natural language explanations.
  • Incorporated expert knowledge through a human-in-the-loop component to enhance interpretability.
  • Evaluated performance of various large language models using rubric-driven assessments.
  • Smaller models showed improved explanatory performance when provided with contextual grounding.
  • The gap in performance between smaller and larger models narrowed significantly.
  • The approach maintained lower computational demands while enhancing relevance for security analysts.

Cite This Study

Padín-Torrente et al. (2026) studied this question.

synapsesocial.com/papers/69db37964fe01fead37c5a4chttps://doi.org/10.1007/s10796-026-10717-3
View Full Paper
Ask AI
Bookmark
Share

Also Consider

Synapse has enriched 5 closely related papers on similar clinical questions. Consider them for comparative context:

  1. 1Enhancing Machine Learning Model Interpretability in Intrusion Detection Systems through SHAP Explanations and LLM-Generated Descriptions2024 · 26 citations
  2. 2Bridging trust and performance in intelligent systems: Hybrid explainable AI approaches for interpreting large language models2026
  3. 3Toward practical human-interpretable explanations2025
  4. 4An explainable and latency-aware unified framework for hybrid graph cyberattack detection2026
  5. 5Applied Explainability for Large Language Models: A Comparative Study2026