Windows 10 updates are more efficient thanks to machine learning, among others. This is the conclusion from one of Microsoft's latest publications, in which the manufacturer explains how important it is to him to care for the quality of the entire process.
The problem of the questionable quality of the Windows 10 update was seriously analyzed for the first time this year. The October 2018 update from the end of 2018 has only reached a small proportion of Windows 10 users by a number of issues, including the most serious bug removing files from disks. Microsoft paused the update, and then tested it again for several weeks with the help of insiders to finally resume its implementation on a slightly different basis.
The May 10 Windows update was to be the first to focus on refining the details. The manufacturer admitted that he did not approach the topic of testing with due diligence, and some testers' comments were not noticed by programmers. As a result, a few months later the Feedback Hub interface was refreshed.
Machine learning helps in updating Windows 10
Now Microsoft is explaining the importance of machine learning in the process of implementing the Windows 10 update. There are many factors to be analyzed that affect the effectiveness of the entire operation, including numerous hardware configurations on users' computers. The source of problems after the upgrade is often incompatibility with third-party software or troublesome drivers.
From the publication we learn briefly how this problem was solved. Users who install updates themselves without waiting for a proposal from the system are of key importance here. Each such installation is checked in detail to develop rules based on which a given Windows 10 update can be offered to other people.
In other words, machine-learning algorithms examine the configuration and version of the system downloaded by the user, and if the process was successful, the chances of proposing updates to other people with similar equipment increase. Microsoft explains briefly that at this stage, among other things, the characteristics of components, drivers and applications are analyzed.
The use of machine learning in the error prediction process is also a relatively new approach. Microsoft points out that previously all problems in Windows were detected in different ways, but to a large extent involving employees. We are talking about the feedback center, telephone technical support and laboratory tests, among others.
Here, Microsoft boasts of using Microsoft Azure Databricks and identifying problems that occur in a large number of hardware. On this basis, the update is automatically suspended on other equipment with similar features.
Windows 10 1909 behind the belt
All this explanation, although it appeared on the network a few days ago, is probably an introduction to the next system update. Windows 10 19H2 is another version that was supposed to be presented together with new Surface series devices.
Unfortunately this did not happen. Instead, we received a show of many new devices and a preview of Windows 10X, which will appear on the market with Microsoft Surface Neo – dual-screen hardware that resembles both a tablet and a laptop. Its premiere is scheduled for next year's vacation.