iERP.ai Studio platform: Trainer

The trainer section of the platform is designed to train Machine Learning models using customer data locally. 4 simple screens will guide you through the process of data selection, data verification, data configuration, and training.

Trainer: Sources

On this screen, you can select sources of the data which will be used for training of algorithm. A list of sources is dependent and pre-configured for each algorithm specifically.

Sources could be in the form of CSV files or direct database/API integration into a specific ERP system.

This configuration is stored in the project file trainer could be triggered automatically to reload data from specified sources.

Trainer: Columns and verification

On this screen you will be able to customize mapping between source data <-> expected internal fields as well as verify the data quality of the data.

Customise mapping: There may be cases where you need to change source column names in order to correctly process the data.

Verification of data quality: For each algorithm we prepared a set of rules which are designed to verify that data quality is sufficient to execute Machine learning. You may see warnings that will allow you to progress to the next step or errors which require correction before training could take place.

PDF data quality report: You are able to generate detailed data quality PDF reports which will help you understand what exactly needs to be corrected.

Trainer: Data

On this screen, you see and browse data loaded into the system, add records and configure data for the trainer.

Data: During implementation this screen will give you peace of mind that data has been correctly loaded. When you are testing algorithms, there may be a need to add or correct training records and this can be accomplished on this screen.

Configuration of data for trainer: Depending on the algorithm, a different configuration of data could be done. For example, selecting which records are dedicated to training and which for testing. More information is available in the consultant handbook.

Trainer: Training

On this screen, you execute training of AI/ML algorithms using customer local data.

Training configuration: Each algorithm and specific customer data does require specific configuration. We are providing detailed recommendations on how to set up the trainer to achieve optimal results.

Training execution: Training is executed on this screen and it will take anywhere from several seconds to several hours. This is highly dependent on a number of records that will be processed by trainer and performance on a machine where the learning is executed.

Scheduling of training: Platform will allow you to schedule training to run automatically. This will allow the algorithm to adapt even when your business and/or data will change.