Launched earlier this year, Lobe aims to make deep learning simple. Nicknamed “the machine learning platform for complete idiots,” the software allows users to build custom deep learning models, train them and then use the models in-app — without any code.
Excited to finally show all of you what we have been working on for the past two years! Check out @lobe_ai to quickly get started with deep learning. We strongly believe shortening the feedback loop with AI and making it understandable will help push research/industry forward May 2, 2018
The company is venture-backed and launched after two years of development.
Lobe doesn’t require installation or experience, and it runs on any platform. To start using it, simply go to Lobe.ai and sign in with a Google account. According to The Verge, Lobe wasn’t built in an attempt to compete with pro tools, rather it was created to help amateurs make their way in.
I'm blown away by how simple @Lobe_AI is making the development and use of custom deep learning models. You should absolutely watch this 10min video to understand how it works (and the complexities related to #AI training): May 3, 2018
Lobe works by separating the different parts of the neural network into boxes, or “lobes.” Users can access these “lobes,” which are essentially pre-made plugins and and alter how each one processes data. The process allows users to build their projects off of intelligence that’s already been created. The final product can be exported to platforms including TensorFlow by Google and the iOS-based CoreML by Apple.
The web application is currently in beta, but some of the projects that have already been built using it are pretty impressive. One of the featured projects in the video was an algorithm that can calculate a specific angle from the position of someone’s hand.
With the entrance of this tool that makes it much easier for those not very well-versed in programming to start building deep learning models, what kind of direction does the profession seem to be taking?
“I think the AMOUNT of custom programming the typical company will do themselves will drop, and this will also affect deep learning programming (e.g. see tools like @lobe_ai for drag and drop DL),” wrote Julian Harris, the head of technology research at CognitionX, on Twitter.