Yet how far it could go is being held back, in some ways, by a shortage of engineers and scientists to work on it. Which has some folks asking, what if there was another way to develop the technology?
In fact, what if machine learning software could be developed by machine learning software?
That’s exactly what Google is working on, and the implications are tremendous for enterprises eager to deploy AI and machine learning to improve operational efficiency, employee productivity and customer service.
Google’s initiative, called AutoML, was a prime topic at the company’s recent I/O software developer conference, and was also explained in a recent blog post by Quoc Le & Barret Zoph, research scientists on the Google Brain AI research group.
“Typically, our machine learning models are painstakingly designed by a team of engineers and scientists,” they write. Thus the “process of designing networks often takes a significant amount of time and experimentation by those with significant machine learning expertise.”
AutoML is an effort to bypass this human-based process. The goal is to automate the design of machine learning models through reinforcement-learning algorithms. According to the Google research scientists:
“In our approach, a controller neural net can propose a “child” model architecture, which can then be trained and evaluated for quality on a particular task. That feedback is then used to inform the controller how to improve its proposals for the next round. We repeat this process thousands of times — generating new architectures, testing them, and giving that feedback to the controller to learn from. Eventually the controller learns to assign high probability to areas of architecture space that achieve better accuracy on a held-out validation dataset, and low probability to areas of architecture space that score poorly.”
Early experiments with AutoML have been promising. AI-created architectures to solve natural language and image-recognition problems have rivaled or surpassed those of human engineers and scientists.
“Perhaps more significantly,” writes Tom Simonite in MIT Technology Review, AutoML “came up with architectures of a kind that researchers didn’t previously consider suited to those tasks. ‘In a sense it found something we didn’t know about,’ says Le. ‘It’s striking.’”
Le and Zoph believe that automating AI and deep learning for specific tasks will democratize these processes and make it easier to accelerate their impact in the enterprise and beyond.
“If we succeed, we think this can inspire new types of neural nets and make it possible for non-experts to create neural nets tailored to their particular needs, allowing machine learning to have a greater impact to everyone,” they write.
This post first appeared on DXC.Technology in May 2017.
Author: Chris Nerney
Chris Nerney is a technology writer who covers mobile technology, big data and analytics, Android, data centers and cloud computing. He lives in upstate New York