Datasets

The Problem with Today’s NIDS Datasets

Most network intrusion detection research relies on datasets collected in controlled lab environments or generated synthetically. These datasets contain “design smells” — accidental patterns that classifiers exploit to achieve high accuracy in the lab but that don’t hold up in deployment. A classifier might learn to flag attacks based on packet timing artifacts from the lab network rather than the actual malicious payload.

What You’ll Work On

In this theme, you’ll help build datasets that bridge the gap between lab and reality, working with TDC NET to validate against real network conditions.

Possible thesis directions:

What You’ll Learn

Project Results

NIDS-Verify

NIDS-Verify combines adversarial training with formal verification for network intrusion detection: attack patterns are formally specified in the Vehicle language and enforced during training through projected gradient descent over hyperrectangle-bounded input regions. Extending the companion code of Flood et al., 2024, Gusts Grinbergs re-engineered the dataset and verification pipeline — rebuilding feature extraction and preprocessing for CIC-IDS-2017, adding tooling for large-scale packet captures, and correcting known timestamp inconsistencies in the dataset — and modernised the training stack through a migration to PyTorch Lightning and integration with the property-driven-ml framework.

Relevant Literature

Supervisors: Alessandro Bruni (ITU), Nicola Dragoni (DTU) – see Team