Google open-sources NLS (Neural Structured Learning), a machine learning framework to train neural networks with structured data and graphs.
Open-source software means the source code is released under a license where the rights holder grants anyone the right to study, change, and distribute the software. This is huge! Potentially letting more people play in the machine learning arena.
Being open-source isn’t the only reason this is a big deal. AI models often process are unstructured like images, video, and audio. But structured file types are useful for machine learning projects to train AI’s in pattern recognition. This could possibly be applied to medical data like MRI or X-ray.
NSL also let’s non developers play with AI, machine learning. As the framework requires very little code, in some cases only five lines to get something running.
This framework also let’s people create “adversarial examples: files that have manipulated pixels, that can corrupt an AI’s processing. Letting users build machine learning models that could fight off possible attacks.
It may be that AI, like this will lead to better human experience and longer lives thanks to medical advancement, or this could be the first step toward training a T-1000, but this time without Dutch around to save us all.