Microsoft has released a free AI training tool for those who cannot program

Wednesday, October 28, 2020, 7:49 GMT

Microsoft has opened up access to the Lobe program globally, allowing you to build ready-made machine learning models for further use of third-party apps and applications in a matter of minutes. In the process of training models, the algorithm is totally free and excludes the writing of program code-users simply need to load data into it, and it will do all the work itself.

Achine for dummies learning

For training machine learning models, Microsoft has launched Lobe. It is targeted at the largest possible mass of users, as no programming skills and abilities are needed at all. That is, with its assistance, anyone, even if not related to the IT industry, can train artificial intelligence. In just 10 minutes, Lobe enables you to build a machine learning model from scratch, according to the developers.

Lobe is a standalone application that is run on a device or laptop and does not require cloud resources to be linked further. Lobe explicitly executes all calculations on the user’s computer, according to Microsoft.

Lobe was only released in beta at the time of release, and only for Windows and MacOS.

Developers do not report the date of the launching of the final version of the service and its appearance on other devices, such as those for Android and iOS.

Why Lobe Functions

In fact, the beta version of Lobe has minimal functionality, showing the program ‘s promise. At the time of the beta update, it just learned how to identify pictures automatically. The user needs to load the pre-prepared files into the program, then mark them, and send a ready-made model to the program at the output. Microsoft noted that the accuracy of image recognition will not be sufficiently high. By retraining the software, the user will be able to improve it individually.

The whole Lobe interface is built on the concept of drag-and – drop. The software can process the data after the user passes it to the software, and the final model will then be unloaded, for example, using the CoreML and TensorFlow systems, into the appropriate application for iOS and Android. It is also possible to directly export to the cloud-Microsoft Azure, Google Cloud and Amazon Web Sites are supported.

Several ways of using Lobe were given by the developers. You will train machine learning models using this service, which are then used to examine aerial and animal and plant images, recognize masks on the faces of humans, recognise movements and emotions, and even forms of sports exercises.

With Lope, you can build smoke and fire detection models that can be useful for detecting forest fires at an early level , for example. Another use case is developing a model for smart baby monitors for machine learning. If the infant is unconscious or already awake, they will be able to tell him and alert the parents of his waking until he begins to scream.

How it all started
From scratch, Microsoft didn’t grow Lobe. It was originally a separate start-up business of the same name, formed in San Francisco, USA, in 2016, according to CrunchBase. The roots are Adam Menges, Markus Beissinger and Mike Matas.

The only project the start-up team was focused on was Lobe. Their aim was to make available to all users a software for training machine learning models.

In 2018 , Microsoft became involved in Lobe. The start-up became its property in September 2018, but the financial and other terms of the transaction were not revealed by the parties. The founders of Lobe only said in their message that after the transition to Microsoft, the Lobe service will remain an independent app.

Honeycode is absolutely free to download, and both ordinary users and major developers have access to it. In particular, in the production of new iterations of the corporate messenger Slack, the features of this service will be included.

A special graphical interface with a set of special models for designing programs of different kinds is introduced by Honeycode. They are case management, budget acceptance, inventory manager, CMS, content tracker, and so on, for example. Amazon itself implemented a job manager (Simple To-do) and a method for analyzing the work of company workers for customers (CRM, Customer Tracker) as an example of apps created using Honeycode.