{
  "id": 1198108,
  "title": "A reproducible evaluation framework for benchmarking machine learning and hybrid ensemble models in power system anomaly detection",
  "url": "https://urgent.news/2026/08/16/a-reproducible-evaluation-framework-for-benchmarking-machine-learning",
  "topic": "ai",
  "section": "AI",
  "published": "2026-08-16T00:00:00.000Z",
  "source": {
    "name": "Scientific Reports",
    "slug": "scientific-reports",
    "url": "https://www.nature.com/articles/s41598-026-67295-2"
  },
  "original_language": "en",
  "account": "Power system anomaly detection faces heightened risks due to digitalization and cyber threats. This paper introduces a reproducible evaluation framework to benchmark machine learning and hybrid ensemble models. Five machine learning models and two hybrid ensemble approaches are compared using voltage, current, load, frequency, power factor, and Total Harmonic Distortion measurements. Gradient Boosting achieves the highest accuracy of 0.9962, while hybrid models reach accuracies of 0.9989 (Voting Hybrid) and 0.9994 (Stacking Hybrid). The framework incorporates SHapley Additive exPlanations (SHAP) and Local Interpretable Model-Agnostic Explanations (LIME) for interpretability. The study's main contribution is providing a unified, reproducible evaluation framework for fair comparison and systematic benchmarking of anomaly detection models in cyber-physical power systems.",
  "summary": "Scientific Reports, Published online: 16 August 2026; doi:10.1038/s41598-026-67295-2 A reproducible evaluation framework for benchmarking machine learning and hybrid ensemble models in power system anomaly detection",
  "key_points": [],
  "editors_take": null,
  "illustration": null,
  "coverage": {
    "outlets": 1,
    "also_reported_by": []
  },
  "ai_generated": true,
  "disclaimer": "Summaries, key points and the editor’s take are written by software from other outlets’ reporting and may contain errors — always check the linked original."
}