Google Is Testing A Technique To Learn From User' Data While Protecting Privacy

Latest updates have revealed that Google is testing a new way to learn from the data of users while protecting their privacy at the same time. According to reports, the company has outlined a new machine learning model, which it is testing in an attempt to increase the protection of user privacy.

It is reported that Google has already discussed the new process in a report that was posted on its research blog last week. In the blog post, the company stated that the procedure is based on a 2012 suggestion from the White House. It is believed that the procedure will further limit the amount of data that is collected by companies.

Google claims that the new process which is described as Federated Learning, can aid in the process of data collection and training on a user's device and that this learning can still be shared. According to Android Authority, training algorithms generally requires the storage of user data on servers so as to process it. But the possibilities of potential security threat can not be overruled, as the report noted that cloud-based data could be the target of hackers.

However, it is reported that the process does not mean that some of the user's data is transferred to Google's servers, but it involves an encrypted summary that is mixed with the data of other users in order to anonymize it. The developer noted that users' original content is never taken from their devices.

According to Quartz, Google requires the data of users on its servers so that its AI will learn properly. But the new technique now gives Google the ability to deputize users' devices thereby learning from their actions. The technique then allows a summary of what is learned to be sent to Google servers.

The report also stated that Google is currently testing the Federated Learning technique through its Gboard keyboard on Android. It is reported that this is to enable the developer to learn how individuals utilize the service.

"When Gboard shows a suggested query, your phone locally stores information about the current context and whether you clicked the suggestion. Federated Learning processes that history on-device to suggest improvements to the next iteration of Gboard's query suggestion model, "Google stated in the blog post.

However, it is said that the benefit doES not just relate to security, but also extend to speed as the Gboard is capable of using what is been learned even before it receives an update from Google.