Stage 6

Development and research of methods and algorithms for training composite models with multilevel nesting; and development of effective tools for their application in the framework.
Development and study of methods and algorithms for pre-learning and transfer learning of composite models with partial structure replacement, hyperparameter tuning and basic updating of weights of atomic models.
Study the effectiveness of algorithms for learning the structure of composite models and develop alternative hybrid algorithms based on reinforcement learning methods, Bayesian optimization, etc.
Developing a semi-automated process for updating framework results on popular ML benchmarks, and testing the framework on Kaggle competition data.
Developing a set of FEDOT use cases for different subject areas: oil, geo-data, finance, etc.
Support for enterprise formats of input data
Integration of the FEDOT framework with ML lifecycle management platforms (e.g. MLflow).

Assign some issues to this milestone.