Anomaly detection metrics analyzed for imbalanced datasets

This research paper delves into the complexities of evaluating anomaly detection models, particularly when faced with significant class imbalance. The authors analyze the behavior of common metrics like AUROC, AUPR, F1-score, and MCC across varying levels of imbalance. They introduce the concept of metric landscapes, which visually represent how metric values relate to true positive and true negative rates, offering a clearer understanding of metric preferences and stability. The findings aim to provide practical guidance for researchers and practitioners in interpreting and comparing anomaly detection results across datasets with different imbalance ratios. AI

IMPACT Provides guidance on interpreting anomaly detection results, crucial for developing more robust AI systems in imbalanced data scenarios.

RANK_REASON Academic paper analyzing evaluation metrics for anomaly detection. [lever_c_demoted from research: ic=1 ai=1.0]

Read on arXiv stat.ML →

• alphaXiv
• Anomaly Detection
• arXiv
• AUROC
• CatalyzeX
• DagsHub
• F1-score
• Gotit.pub
• Hugging Face
• IArxiv
• Influence Flower
• ScienceCast

• paper
• other

AI-generated summary · Google Gemini · from 1 sources. How we write summaries →