Time-Series Anomaly Detection Benchmark

This publish branch is a reduced benchmark package for kube-watch-v2 experiments.

Included Models

• OCSVM
• IForest
• LOF
• DAGMM
• DeepSVDD
• THOC
• LSTM-AD
• Anomaly Transformer
• DCdetector
• OmniAnomaly
• USAD

Layout

Code/
  main.py
  solver.py
  config/ocsvm.json
  data_factory/kube_watch.py
  model/
  scripts/run_ocsvm_kube_watch_v2.sh
Datasets/kube-watch-v2/
  node_30d/stack/
  kube_pod_10d/stack/
  kube_controlplane/stack/
result/

Dataset directories are placeholders only. Put kube-watch-v2 arrays and segment files under the matching dataset path before running.

Run

pip install -r requirements.txt
bash Code/scripts/run_ocsvm_kube_watch_v2.sh