tribuo

Why we included this project

If your team builds on the JVM, Tribuo is worth a look: it's a Java-native machine learning library that covers classification, regression, clustering, anomaly detection, and multi-label prediction through one consistent API, and it wraps popular engines like XGBoost and liblinear so you don't have to maintain those integrations yourself. Training is configured through Oracle's OLCUT system, meaning you can describe a trainer in XML or JSON and rebuild the same model repeatedly, which suits teams that want reproducible pipelines. Every model and evaluation carries a serializable provenance object recording the data, transformations, and hyperparameters behind it, and that provenance can be hashed for tracking through an external system. Many models also export to ONNX, so you can train in Java and serve the result from other languages or cloud platforms.