Because Google issued several paper on federated learning on mobile device, like Gboard, but I can not find docs on training on TFLite website, does TFLite support on device training at now?
Deploy machine learning models on mobile and edge devicesExplore TensorFlow Lite Android and iOS apps.
In summary, if you can use tensorflow lite, I use it daily in Windows, MacOS, and Linux, it is not necessary to use Docker at all. Just a python file and that's it.
Multiple platform support, covering Android and iOS devices, embedded Linux, and microcontrollers. Diverse language support, which includes Java, Swift, Objective-C, C++, and Python.
We do not support general training (or federated training) on TensorFlow Lite. We however have released a tutorial on doing on-device personalization with TF lite. https://blog.tensorflow.org/2019/12/example-on-device-model-personalization.html
It basically goes through the process of exporting subgraphs that do training substeps and evaluation to stage training efforts manually. Personalization is the most practical on-device training mechanism right now, so hopefully this addresses your needed.
We are also working on a more comprehensive training solution that will eventually support things like federated learning and also more easy-to-use workflows.
If you love us? You can donate to us via Paypal or buy me a coffee so we can maintain and grow! Thank you!
Donate Us With