AI Without Coding
Microsoft offers free AI-building tool that requires no programming expertise to train custom machine learning models
Amidst the ever-increasing use of Artificial Intelligence (AI) in all possible technologies, Microsoft is working on an application that would make creation of AI models a breeze – no coding or software programming expertise required! Named Lobe, the app would allow anyone to train a custom machine learning model without writing a single line of programming code. And what’s more, it is absolutely free.
Microsoft has recently released a public preview of the application. Available as a desktop app for Windows and Mac, Lobe only supports image classification as of now. However, Microsoft assures that new releases to support other neural network models and data types are coming soon.
Being code-free and image-only for now doesn’t diminish the effectiveness of Lobe by any count. It has already been put to good use by a number of customers. Nature Conservancy – an NGO that deals with environmental issues is one of them; Sincro– a Seattle-based automobile marketing firm is another. According to Microsoft statements, these early customers have used the current version of Lobe to create AI applications that perform a wide range of activities: tracking tourist activity around Caribbean coral reefs, identifying harmful plants or beehive invaders like wasps, alerting people when their garage door is open, or scanning online ads to filter out unattractive car images.
So how does Lobe work? The company succinctly sums it up on their website: “Lobe simplifies the process of machine learning into three easy steps. Collect and label your images. Train and understand your results. Then play with your model and improve it.” Sounds absurdly easy, but that is how Microsoft intends the application to be.
To create an AI in Lobe, a user first needs to import and label a collection of images of what they want Lobe to recognize. This collection would be the dataset used to train the application. Lobe will then analyse the images and select the most suitable open-source machine learning architecture for the input dataset. This is done by sifting through a built-in library of neural network architecture. Finally, the application will start training the model on the user’s device based on the provided data – creating an AI model optimised to scan images for the object or action as specified by the user. In addition, the user can review the model’s performance through real-time visual results, provide feedback on the output, and rectify erroneous labels.
To quote the company website yet again, the entire process can be simply summed up as follows: “Just show it examples of what you want it to learn, and it automatically trains a custom machine learning model that can be shipped in your app.” And all of this happens on your own computer without any setup or configuration, and without uploading anything to the Cloud – which means your data remains secure and private throughout. After training, the model can be exported to a variety of industry-standard formats and shipped on a platform of the user’s choice.
Lobe has its parallel in AutoML – a technology that can automate parts and most of the machine learning creation workflow, reducing development effort and cost. Microsoft has already made available AutoML features in its Azure public Cloud for enterprises to use. However, AI capabilities in Azure are exclusively for advanced projects. In contrast, Lobe supports even small-scale and uncomplicated requirements by amateurs – something which no existing AI tool offers.
The Lobe website makes it amply clear that Microsoft will not skimp on features. It offers a lot of options even in this image-only version; and that makes it clear that the full version would be feature-rich as well. As promised in their website: “Lobe has everything you need to bring your machine learning ideas to life.”
Anyone interested can download and try out Lobe from either of the following links:
A video tutorial for Lobe is also available on YouTube: https://www.youtube.com/watch?v=Mdcw3Sb98DA&feature=emb_title
