Overview (Tribuo 4.1.1 API)

Core Packages
Package	Description
org.tribuo	Provides the core interfaces and classes for using Tribuo.
org.tribuo.data	Provides classes for loading in data from disk, processing it into examples, and splitting datasets for things like cross-validation and train-test splits.
org.tribuo.data.columnar	Provides classes for processing columnar data and generating `Example`s.
org.tribuo.data.columnar.extractors	Provides implementations of `FieldExtractor`.
org.tribuo.data.columnar.processors.feature	Provides implementations of `FeatureProcessor`.
org.tribuo.data.columnar.processors.field	Provides implementations of `FieldProcessor`.
org.tribuo.data.columnar.processors.response	Provides implementations of `ResponseProcessor`.
org.tribuo.data.csv	Provides classes which can load columnar data (using a `RowProcessor`) from a CSV (or other character delimited format) file.
org.tribuo.data.sql	Provides classes which can load columnar data (using a `RowProcessor`) from a SQL source.
org.tribuo.data.text	Provides interfaces for converting text inputs into `Feature`s and `Example`s.
org.tribuo.data.text.impl	Provides implementations of text data processors.
org.tribuo.dataset	Provides utility datasets which subsample or otherwise transform the wrapped dataset.
org.tribuo.datasource	Simple data sources for ingesting or aggregating data.
org.tribuo.ensemble	Provides an interface for model prediction combinations, two base classes for ensemble models, a base class for ensemble excuses, and a Bagging implementation.
org.tribuo.evaluation	Evaluation base classes, along with code for train/test splits and cross validation.
org.tribuo.evaluation.metrics	This package contains the infrastructure classes for building evaluation metrics.
org.tribuo.hash	Provides the base interface and implementations of the `Model` hashing which obscures the feature names stored in a model.
org.tribuo.impl	Provides implementations of base classes and interfaces from `org.tribuo`.
org.tribuo.json	Provides interop with JSON formatted data, along with tools for interacting with JSON provenance objects.
org.tribuo.math	Contains the implementation of Tribuo's math library, it's gradient descent optimisers, kernels and a set of math related utils.
org.tribuo.math.kernel	Provides a `Kernel` interface for Mercer kernels, along with implementations of standard kernels.
org.tribuo.math.la	Provides a linear algebra system used for numerical operations in Tribuo.
org.tribuo.math.optimisers	Provides implementations of `StochasticGradientOptimiser`.
org.tribuo.math.optimisers.util	Provides some utility tensors for use in gradient optimisers.
org.tribuo.math.util	Provides math related util classes.
org.tribuo.provenance	Provides Tribuo specific infrastructure for the `Provenance` system which tracks models and datasets.
org.tribuo.provenance.impl	Provides internal implementations for empty provenance classes and TrainerProvenance.
org.tribuo.sequence	Provides core classes for working with sequences of `Example`s.
org.tribuo.transform	Provides infrastructure for applying transformations to a `Dataset`.
org.tribuo.transform.transformations	Provides implementations of standard transformations like binning, scaling, taking logs and exponents.
org.tribuo.util	Provides utilities which don't have other Tribuo dependencies.

Shared Backend Packages
Package	Description
org.tribuo.common.liblinear	Provides base classes for using liblinear from Tribuo.
org.tribuo.common.libsvm	The base interface to LibSVM.
org.tribuo.common.nearest	Provides a K-Nearest Neighbours implementation which works across all Tribuo `Output` types.
org.tribuo.common.sgd	Provides the base classes for models trained with stochastic gradient descent.
org.tribuo.common.tree	Provides common functionality for building decision trees, irrespective of the predicted `Output`.
org.tribuo.common.tree.impl	Provides internal implementation classes for building decision trees.
org.tribuo.common.xgboost	Provides abstract classes for interfacing with XGBoost abstracting away all the `Output` dependent parts.

Anomaly Detection Packages
Package	Description
org.tribuo.anomaly	Provides classes and infrastructure for anomaly detection problems.
org.tribuo.anomaly.evaluation	Evaluation classes for anomaly detection.
org.tribuo.anomaly.example	Provides a anomaly data generator used for testing implementations.
org.tribuo.anomaly.liblinear	Provides an interface to LibLinear-java for anomaly detection problems.
org.tribuo.anomaly.libsvm	Provides an interface to LibSVM for anomaly detection problems.

Multi-class Classification Packages
Package	Description
org.tribuo.classification	Provides classes and infrastructure for multiclass classification problems.
org.tribuo.classification.baseline	Provides simple baseline multiclass classifiers.
org.tribuo.classification.dtree	Provides implementations of decision trees for classification problems.
org.tribuo.classification.dtree.impl	Provides internal implementation classes for classification decision trees.
org.tribuo.classification.dtree.impurity	Provides classification impurity metrics for decision trees.
org.tribuo.classification.ensemble	Provides majority vote ensemble combiners for classification along with an implementation of multiclass Adaboost.
org.tribuo.classification.evaluation	Evaluation classes for multi-class classification.
org.tribuo.classification.example	Provides a multiclass data generator used for testing implementations.
org.tribuo.classification.experiments	Provides a set of main methods for interacting with classification tasks.
org.tribuo.classification.explanations	Provides core infrastructure for local model based explanations.
org.tribuo.classification.explanations.lime	Provides an implementation of LIME (Locally Interpretable Model Explanations).
org.tribuo.classification.liblinear	Provides an interface to LibLinear-java for classification problems.
org.tribuo.classification.libsvm	Provides an interface to LibSVM for classification problems.
org.tribuo.classification.mnb	Provides an implementation of multinomial naive bayes (i.e., naive bayes for non-negative count data).
org.tribuo.classification.sequence	Provides infrastructure for `SequenceModel`s which emit `Label`s at each step of the sequence.
org.tribuo.classification.sequence.example	Provides a classification sequence data generator for smoke testing implementations.
org.tribuo.classification.sequence.viterbi	Provides an implementation of Viterbi for generating structured outputs, which can sit on top of any `Label` based classification model.
org.tribuo.classification.sgd	Provides infrastructure for Stochastic Gradient Descent for classification problems.
org.tribuo.classification.sgd.crf	Provides an implementation of a linear chain CRF trained using Stochastic Gradient Descent.
org.tribuo.classification.sgd.kernel	Provides a SGD implementation of a Kernel SVM using the Pegasos algorithm.
org.tribuo.classification.sgd.linear	Provides an implementation of a classification linear model using Stochastic Gradient Descent.
org.tribuo.classification.sgd.objectives	Provides classification loss functions for Stochastic Gradient Descent.
org.tribuo.classification.xgboost	Provides an interface to XGBoost for classification problems.

Clustering Packages
Package	Description
org.tribuo.clustering	Provides classes and infrastructure for working with clustering problems.
org.tribuo.clustering.evaluation	Evaluation classes for clustering.
org.tribuo.clustering.example	Provides a clustering data generator used for testing implementations.
org.tribuo.clustering.kmeans	Provides a multithreaded implementation of K-Means, with a configurable distance function.

Multi-label Classification Packages
Package	Description
org.tribuo.multilabel	Provides classes and infrastructure for working with multi-label classification problems.
org.tribuo.multilabel.baseline	Provides an implementation of independent multi-label classification that wraps a `Label` `Trainer` and uses it to make independent predictions of each label.
org.tribuo.multilabel.evaluation	Evaluation classes for multi-label classification using `MultiLabel`.
org.tribuo.multilabel.example	Provides a multi-label data generator for testing implementations.
org.tribuo.multilabel.sgd	Provides infrastructure for Stochastic Gradient Descent for multi-label classification problems.
org.tribuo.multilabel.sgd.linear	Provides an implementation of a multi-label classification linear model using Stochastic Gradient Descent.
org.tribuo.multilabel.sgd.objectives	Provides multi-label classification loss functions for Stochastic Gradient Descent.

Regression Packages
Package	Description
org.tribuo.regression	Provides classes and infrastructure for regression problems with single or multiple output dimensions.
org.tribuo.regression.baseline	Provides simple baseline regression predictors.
org.tribuo.regression.ensemble	Provides `EnsembleCombiner` implementations for working with multi-output regression problems.
org.tribuo.regression.evaluation	Evaluation classes for single or multi-dimensional regression.
org.tribuo.regression.example	Provides some example regression data generators for testing implementations.
org.tribuo.regression.impl	Provides skeletal implementations of `Regressor` `Trainer` that can wrap a single dimension trainer/model and produce one prediction per dimension independently.
org.tribuo.regression.liblinear	Provides an interface to liblinear for regression problems.
org.tribuo.regression.libsvm	Provides an interface to LibSVM for regression problems.
org.tribuo.regression.rtree	Provides an implementation of decision trees for regression problems.
org.tribuo.regression.rtree.impl	Provides internal implementation classes for the regression trees.
org.tribuo.regression.rtree.impurity	Provides implementations of regression tree impurity metrics.
org.tribuo.regression.sgd	Provides infrastructure for Stochastic Gradient Descent based regression models.
org.tribuo.regression.sgd.linear	Provides an implementation of linear regression using Stochastic Gradient Descent.
org.tribuo.regression.sgd.objectives	Provides regression loss functions for Stochastic Gradient Descent.
org.tribuo.regression.slm	Provides implementations of sparse linear regression using various forms of regularisation penalty.
org.tribuo.regression.xgboost	Provides an interface to XGBoost for regression problems.

External Interface Packages
Package	Description
org.tribuo.interop	This package contains the abstract implementation of an external model trained by something outside of Tribuo.
org.tribuo.interop.onnx	This package contains a Tribuo wrapper around the ONNX Runtime.
org.tribuo.interop.onnx.extractors	Provides feature extraction implementations which use ONNX models.
org.tribuo.interop.tensorflow	Provides an interface to TensorFlow, allowing the training of non-sequential models using any supported Tribuo output type.
org.tribuo.interop.tensorflow.example	Example architectures for use with Tribuo's TF interface.
org.tribuo.interop.tensorflow.sequence	Provides an interface for working with TensorFlow sequence models, using Tribuo's `SequenceModel` abstraction.

Utility Packages
Package	Description
org.tribuo.tests	This package provides helper classes for Tribuo's unit tests.
org.tribuo.util.infotheory	This package provides static classes of information theoretic functions.
org.tribuo.util.infotheory.example	This package provides demos for the information theoretic function classes in `org.tribuo.util.infotheory`.
org.tribuo.util.infotheory.impl	This package provides the implementations and helper classes for the information theoretic functions in `org.tribuo.util.infotheory`.
org.tribuo.util.tokens	Core definitions for tokenization.
org.tribuo.util.tokens.impl	Simple fixed rule tokenizers.
org.tribuo.util.tokens.impl.wordpiece	Provides an implementation of a Wordpiece tokenizer which implements to the Tribuo `Tokenizer` API.
org.tribuo.util.tokens.options	OLCUT `Options` implementations which can construct `Tokenizer`s of various types.
org.tribuo.util.tokens.universal	An implementation of a "universal" tokenizer which will split on word boundaries or character boundaries for languages where word boundaries are contextual.

Tribuo is a library for building and deploying Machine Learning models. It provides classification, regression, clustering, anomaly detection and multi-label classification algorithms. Tribuo is strongly typed, with most of the relevant types living in the org.tribuo package. Task specific types live in the root package of that task (e.g., org.tribuo.classification). Tribuo is modular, each package/module is scoped to have minimal dependencies, and there are few cross-cutting packages. Tribuo's development is lead by Oracle Labs' Machine Learning Research Group, the source is hosted on Github, and more documentation is available on the project website tribuo.org.