CICIDS2017: Per-Class Intrusion Detection Performance Metrics

Per-class detection performance metrics for the CICIDS2017 cybersecurity benchmark dataset, as evaluated in a 2026 study. The dataset, authored by Ali Alqazzaz and shared on figshare, likely contains accuracy or other classification metrics for different attack types. It is a small Excel file (5.5 KB) used to validate the MedDefender-MHAN explainable AI model for healthcare IoT security.

Use Cases

• Benchmarking intrusion detection model performance based on per-class metrics mentioned in the description
• Analyzing trade-offs between detection accuracy and model explainability for healthcare IoT security
• Comparing the performance of the proposed MedDefender-MHAN framework against other methods on the CICIDS2017 dataset

Strengths

• Specific detection accuracies (99.47% on CICIDS2017) are cited in the associated research paper
• Explainability metrics (94.6% alignment with expert signatures) are provided for model validation
• Released under a permissive CC-BY-4.0 license

Limitations

• Row count and column-level documentation are absent; field semantics must be inferred after download
• Scale is very limited (5.5 KB), suggesting it contains summary metrics, not the underlying raw data

Provenance

Source

figshare, associated with research on the MedDefender-MHAN model

Collection Method

Likely derived from experimental evaluation results on the CICIDS2017 benchmark

Time Range

Associated research published or updated in 2026

Freshness

Last updated 2026-04-17 17:39:28

Geography

null