Hi everyone!
I’ve published five tabular datasets under BhinnekaAI on Hugging Face, covering different domains and anomaly detection challenges.
Rather than focusing only on obvious outliers, these datasets are designed around more subtle patterns where anomalous samples may remain within normal feature ranges but violate deeper relationships in the data.
1. FRD — Credit Card Fraud Simulator
Explore the dataset Simulates credit card transactions with fraudulent samples hidden within normal feature ranges. The challenge is identifying broken multivariate relationships between transaction amounts, account balances, and daily activity.
2. NET — Network Intrusion Logs
Explore the dataset Represents network traffic with multiple clusters of normal activity. Simulated cyberattacks are placed near narrow gaps between normal clusters, challenging models to detect subtle spatial and local-density anomalies.
3. MED — Patient Biomarker Screening
Explore the dataset Models multivariate blood biomarker profiles with normal covariance structures. Anomalies are represented through changes in the orientation of the underlying multivariate structure.
4. IOT — Sensor Anomaly
Explore the dataset Simulates industrial machine telemetry with nonlinear harmonic patterns. Mechanical anomalies are introduced through phase shifts while keeping individual feature values within their normal amplitude ranges.
5. LOG — Supply Chain Inventory Audit
Explore the dataset Represents high-dimensional logistics records with irrelevant noise. Hidden audit discrepancies, such as inventory inconsistencies and cost inflation, challenge models to identify meaningful anomalies among distracting features.
The goal is to explore how machine learning models handle different types of subtle anomalies across financial, cybersecurity, biological, industrial, and logistical data.
These datasets may be useful for experimenting with unsupervised anomaly detection, comparing baseline algorithms, and studying the limitations of different scoring approaches.
I’d appreciate feedback from the community:
Which dataset presents the most interesting anomaly detection challenge?
Which unsupervised methods would you recommend benchmarking against these datasets?
What additional evaluation scenarios would make this collection more useful for research?
Explore the full collection:
BhinnekaAI (BhinnekaAI) Thanks for checking them out! Feedback and contributions are welcome.
#MachineLearning #AnomalyDetection Datasets #OpenScience