Multi-Output Machine Learning
Predict multiple related labels or numeric targets with chain estimators in browser and Node.js using @kanaries/ml.
View as MarkdownMulti-output models predict several targets for each input row. Independent wrappers are useful when targets do not interact; classifier and regressor chains additionally feed earlier predictions into later members to model target dependencies.
@kanaries/ml exposes these estimators under MultiOutput with serializable, scikit-learn-style APIs for browser and Node.js.
- ClassifierChain for related binary labels.
- RegressorChain for related numeric outputs.
Self-Training Classifier
Pseudo-label confident unlabeled samples with the @kanaries/ml SelfTrainingClassifier JavaScript implementation in browser or Node.js.
ClassifierChain
Model dependencies between multiple binary labels with the @kanaries/ml JavaScript ClassifierChain implementation in browser or Node.js.