Minqi824/ADBench resource

Health v2 · maintenance only

Flags: no_releases

Adoption not part of the score

1021 stars · 152 forks observed · 2026-08-28

What it is AI-extracted, prompt v1, taxonomy v1, 2026-08-30, confidence not recorded

ADBench is the official implementation of a comprehensive tabular anomaly detection benchmark published at NeurIPS 2022, comparing 30 algorithms across 57 datasets with nearly 100,000 experiments. It is available both as a research codebase and an installable Python package.

Use cases

• benchmark anomaly detection algorithms on tabular data
• compare unsupervised vs semi-supervised vs supervised outlier detection methods
• find datasets for anomaly detection research
• evaluate algorithm robustness under data corruption
• reproduce NeurIPS 2022 anomaly detection benchmark results

When to choose

• you need a rigorous, reproducible comparison of anomaly detection methods
• you want curated tabular datasets with anomaly labels
• you are researching the effect of supervision on outlier detection

When to avoid

• you need anomaly detection for time-series, graphs, or images rather than tabular data
• you just want a production anomaly detection library rather than a benchmark
• you need streaming or real-time anomaly detection

Facets

dataset · maturity active

benchmarking machine-learning data-science machine-learning data-science analytics python cross-platform anomaly-detection outlier-detection tabular-data benchmark semi-supervised-learning neurips-2022

Member repositories

For agents

markdown · JSON · MCP: product_card(name="Minqi824/ADBench")