Ablation Study Results for Network Intrusion Detection Models on Harmonized Data

A 5.5 KB dataset from figshare, last updated on 2026-04-20, containing results from an ablation study on machine learning models for intrusion detection. The work by Shailendra Mishra proposes a unified framework, harmonizing the NSL-KDD and CICIDS2017 datasets and benchmarking models including Random Forest, which achieved 98.0% accuracy and 97.0% F1-score.

Use Cases

• Benchmarking intrusion detection models based on harmonized NSL-KDD and CICIDS2017 datasets.
• Analyzing the impact of feature harmonization on model performance using ablation study results.
• Comparing statistical validation results from tests like Wilcoxon signed-rank and McNemar's.
• Evaluating the proposed cryptographic logging mechanism for experiment reproducibility.

Strengths

• Results include specific performance metrics, such as 98.0% accuracy and 97.0% F1-score for the Random Forest model.
• The framework is designed for reproducibility, incorporating a proposed cryptographic logging mechanism.
• Dataset is licensed under CC-BY-4.0, allowing for open use and modification.

Limitations

• Row count and column-level documentation are unknown, limiting suitability assessment.
• The description notes the feature harmonization approach is manual and may not be effective for highly heterogeneous datasets.
• The dataset is very small at 5.5 KB, indicating limited scope, likely containing only summary results.

Provenance

Source

figshare

Collection Method

Results from an experimental framework harmonizing and benchmarking intrusion detection datasets.

Time Range

The study likely uses the NSL-KDD (legacy) and CICIDS2017 (contemporary) datasets, but specific temporal coverage is not stated.

Freshness

Last updated 2026-04-20 17:36:50; freshness should be verified.

Geography

Network traffic data; geographic coverage is not specified.