Before we take a deep dive into TensorFlow Lite, let's try to understand what are the advantages of doing ML on edge devices like mobile/tablet and others.
- Privacy: If inference on a ML model can be performed on a device, user data doesn't need to leave the device, which helps in preserving the privacy of the user.
- Offline predictions: The device doesn't need to be connected to a network to make predictions on a ML model. This unlocks a lot of use cases in developing nations such as India where network connectivity is not so great.
- Smart devices: This can also enable the development of smart home devices such as microwaves and thermostats with on-device intelligence.
- Power efficient: An on-device ML can be more power-efficient as there is no need to transfer data back and forth to the server.
- Sensor data utilization: ...
