As consumers care more and more about their privacy, companies are devising techniques to run deep learning models while respecting this demand.
Let's use a large-scale example from Apple. When browsing through photos on an iOS device, you may notice that it is possible to search for objects or things—cat, bottle, car will return the corresponding images. This is the case even if the pictures are not sent to the cloud. For Apple, it was important to make that feature available while respecting the privacy of its users. Sending pictures for processing without the users' consent would have been impossible.
Therefore, Apple decided to use on-device ML. Every night, when the phone is charging, a computer vision model is run on the iPhone to detect objects in the image and make this feature available.
