Network Intrusion Detection Benchmark Data for DRL Validation

Khorshed Alam's work compares intrusion detection models using deep reinforcement learning and data balancing techniques. The dataset likely contains processed network flow data from multiple benchmark sources like NF-BoT-IoT and NF-CSE-CIC-IDS2018. The source code and supporting data were published on figshare in April 2026.

Use Cases

• Benchmarking intrusion detection algorithms based on network flow data
• Evaluating data balancing techniques like K-means SMOTE for handling class imbalance
• Testing adaptive deep reinforcement learning models on unseen attack types
• Validating model robustness across different network environments using multiple benchmark datasets

Strengths

• Multi-dataset validation across six named benchmark sources ensures robustness
• Explicit evaluation of four advanced data balancing techniques (Borderline-SMOTE, SMOTE-ENN, ADYSN, K-means SMOTE)
• Source code is publicly available on GitHub for reproducibility

Limitations

• The dataset is only 5.5 KB, suggesting it contains summary or processed results rather than raw network traffic
• Column-level documentation is absent; field semantics must be inferred after download
• Row count is unknown, which may limit suitability assessment

Provenance

Source

Khorshed Alam

Collection Method

Likely derived from processing and comparing multiple benchmark network intrusion datasets.

Freshness

Last updated 2026-04-16 17:40:30