New tool Fun-TSG generates time series data for anomaly detection evaluation

Researchers have developed Fun-TSG, a new tool designed to generate multivariate time series data for evaluating anomaly detection methods. Existing datasets often lack detailed anomaly labels and insights into data generation processes, hindering the development and comparison of detection models. Fun-TSG addresses this by allowing for both automated and manual generation of time series, providing granular, variable-level anomaly labels and transparency into the underlying dependencies and generative mechanisms. This enables more rigorous performance analysis for a wide range of anomaly detection techniques. AI

IMPACT Enables more robust evaluation of anomaly detection models by providing customizable and interpretable benchmark datasets.

RANK_REASON The item is a research paper detailing a new tool for generating synthetic data for evaluating machine learning models. [lever_c_demoted from research: ic=1 ai=1.0]

Read on arXiv cs.AI →

• alphaXiv
• arXiv
• CatalyzeX Code Finder for Papers
• Connected Papers
• DagsHub
• Fun-TSG
• Gotit.pub
• Hugging Face
• Litmaps
• Pierre LOTTE
• ScienceCast
• scite Smart Citations

• paper
• product
• other

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