BPMN modelling of the training process

On this page, the BPMN diagram depicts the training process of a machine learning model.

There are 2 main BPMN swim lanes, depicting:

1. "Enabling software components" (middle lane) – represents the corresponding asset and the processes, which are undertaken with the asset. The components are deemed to be an irreplaceable part in conducting machine learning-related operations, such as training, development building, evaluation. The components may include software frameworks, libraries, and runtime environments necessary for model training and development.
2. "Machine learning training system" (lower lane) – stands for the collection of tools, services, activities and processes responsible for the training of a machine learning model.

An auxiliary top lane represents an engineer, who jump-starts the training process. The lane does not correspond to any assets within the UML diagram, it is added to depict the starting point of the process.

Overall, an engineer initiates the training process of a machine learning model. Through enabling software components a preprocessed ML model is created, with untrained parameters based on the internally developed design. The internal design is represented with the “intellectual property”. The preprocessed model is transferred to the training system, which has prepared the training data for the training process, including testing data. The testing data plays a key role in testing and refining the model after a training iteration. After training, a trained ML model is produced, it gets deployed after passing defined tests.