Network Intrusion Classification Results for Multiple Benchmark Datasets

Khorshed Alam published results on April 16, 2026, evaluating a novel deep reinforcement learning approach for network intrusion classification. The work tested the method on benchmark datasets NF-BoT-IoT, NF-UNSW-NB15, NF-ToN-IoT, NF-ToN-IoT-v2, NF-CSE-CIC-IDS2018, and NF-UNSW-NB15-v3. The associated data file is 5.5 KB in size and available under a CC-BY-4.0 license.

Use Cases

• Evaluating data balancing techniques like Borderline-SMOTE, SMOTE-ENN, ADYSN, and K-means SMOTE for network intrusion datasets.
• Testing adaptive modeling by excluding specific attack types (e.g., DoS, Backdoor) from training and including them in testing.
• Comparing intrusion detection performance across multiple benchmark network flow datasets.
• Implementing deep reinforcement learning with Echo State Networks for real-time threat detection.

Strengths

• Multi-dataset validation on six benchmark datasets ensures robustness across different network flow data.
• The source code for the proposed method is publicly available on GitHub.
• The file size is 5.5 KB, indicating a compact, focused results dataset.

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 5.5 KB, indicating a very limited scope, likely containing summary results rather than raw traffic data.

Provenance

Source

Khorshed Alam via figshare.

Collection Method

Results from research evaluating a deep reinforcement learning approach for intrusion classification.

Freshness

Last updated 2026-04-16 17:40:23.