Hugging Face's logo

Datasets:
AriaAICompany
/
net-flow-lab

Tasks:
Tabular Classification
Time Series Forecasting
Languages:
English
Persian
Size:
10K<n<100K
Tags:
cybersecurity
network-anomaly-detection
netflow
machine-learning
Synthetic
License:
Dataset card Files
xet
 Community
Dataset Viewer
The dataset could not be loaded because the splits use different data file formats, which is not supported. Read more about the splits configuration. Click for more details.
Couldn't infer the same data file format for all splits. Got {NamedSplit('validation'): (None, {}), NamedSplit('test'): ('json', {})}
Error code:   FileFormatMismatchBetweenSplitsError

Need help to make the dataset viewer work? Make sure to review how to configure the dataset viewer, and open a discussion for direct support.

NetFlow laboratory data-center flows

Synthetic seven-day Zeek/Suricata/CSV fixture pack for the NetFlow anomaly-review demo.

Organization dataset and collection item are public. Live Gradio (alirezaaminzadeh/net-flow) and the organization card (AriaAICompany/net-flow) are queued until the Hugging Face daily Space-creation cap resets. Runnable Space source is stored in demo/. Collection: Aria AI — Cybersecurity.

Hosts are aliases such as web-east-01. Laboratory IPs are RFC1918 or documentation ranges and are aliased before scoring. There is no PCAP in this pack.

Files

  • data/flows.csv
  • data/flows_preview.csv
  • data/gold_events.json
  • data/zeek_conn_sample.log
  • data/suricata_eve_sample.jsonl
  • data/schema.json
  • data/splits.json
  • data/protocol.md
  • data/feature_card.md
  • data/evaluation.json
  • data/sample_preview.json
  • demo/ — runnable Gradio Space source
  • showcase/ — organization static card

Split

Development is 2026-03-01 through 2026-03-04. Test is 2026-03-05 through 2026-03-07. Both dc-east and dc-west appear in each split. Rows that belong to one authored event stay in one split. Train and test review events use different source hosts and rare destinations.

Gold policy

needs_review_* windows are authored ranges for scan-like short multi-peer bursts, beacon-like low-volume rare peers, and exfil-like high outbound bytes to a rare peer. benign_backup and benign_bulk are healthy context. A high byte count alone is not a security-review label.

License

CC-BY-4.0. Keep the synthetic-data label.
Downloads last month
54
Total file size:
4.56 MB

Collection including AriaAICompany/net-flow-lab