This project aims to show an approach to implementing distributed training of a machine learning model on mobile devices.
Swift for TensorFlow is used for creating a pre-trained ML model on shared/proxy data. Then Google Colab and protobuf are used for recreating, making updatable and exporting the pre-trained model in .mlmodel format. The pre-trained model is delivered to the devices with new versions of the app. Core ML 3 is used for on-device re-training on user data, so they do not leave the device; and also for inferencing. Transfer learning and model personalization concepts are used for this process as well.
The implementation process is structured so that only one Swift programming language is used at all stages of working with the model, making this approach even more convenient and reliable thanks to a single code base, i.e. reusing of pre-processing and featurization code in different parts of the distributed system.
This approach can be an alternative to Federated Learning because of its significant difficulties for production use at present, especially in combination with mobile devices.
