[IGNITE-10479] [ML] Umbrella: Ensemble training and inference

• Type: New Feature
• Status: In Progress
• Priority: Major
• Resolution: Unresolved
• Component/s: ml
• Ignite Flags: Docs Required
• Assignee: Alexey Zinoviev
• Reporter: Alexey Zinoviev
• Created: 29/Nov/18 16:10
• Updated: 1 week ago 15:17

Description

We want to unify API/usage of any ensembles of models.
Currently we already have only boosting and bagging and we want to implement stacking.

Stacking is an ensemble learning technique that combines multiple classification or regression models via a meta-classifier or a meta-regressor. The base level models are trained based on a complete training set, then the meta-model is trained on the outputs of the base level model as features.

Issue Links

• is a parent of: IGNITE-9463 [ML] Update ML tutorial with new model composition/update features
• is a parent of: IGNITE-7149 Gradient boosting for decision tree
• is a parent of: IGNITE-10480 [ML] Stacking for training and inference
• is a parent of: IGNITE-10517 [ML] Merge inference and learning architectures
• is a parent of: IGNITE-8867 [ML] Bagging on learning sample
• is a parent of: IGNITE-10573 [ML] Consistent API for Ensemble training
• is a parent of: IGNITE-10575 [ML] Add examples for ensemble training