Component Contributions to Detection Accuracy for APT Simulation

Adel Alshamrani's dataset, last updated June 2026, presents results from a machine learning approach for detecting Advanced Persistent Threats (APTs). The methodology uses the CERT Insider Threat Dataset to simulate APT behaviors via multi-modal data analysis and language models. Experimental results achieved 96.3% detection accuracy and a 42% reduction in false positives compared to state-of-the-art methods.

Use Cases

• Benchmarking APT detection algorithms based on the reported 96.3% accuracy and 42% false positive reduction.
• Simulating APT attack stages using insider threat data patterns as described in the methodology.
• Developing interpretable threat detection systems based on the natural language generation component for explanations.
• Training uncertainty-aware clustering models for threat detection as referenced in the deep evidential clustering component.

Strengths

• Reported experimental results include a specific 96.3% detection accuracy.
• Methodology achieved a quantified 42% reduction in false positives compared to state-of-the-art methods.
• Dataset is openly licensed under CC-BY-4.0.

Limitations

• Row count is unknown, which may limit suitability assessment.
• Column-level documentation is absent; field semantics must be inferred after download.
• The dataset is very small at 5.5 KB, indicating limited scope, likely containing summary results rather than raw behavioral logs.

Provenance

Source

Adel Alshamrani via figshare.

Collection Method

Derived from research applying machine learning to the CERT Insider Threat Dataset to simulate and detect APT behaviors.

Freshness

Last updated 2026-06-02 17:27:34; freshness should be verified.