Artificial intelligence (AI) researchers and companies are applying machine learning to automate the more complicated aspects of AI algorithm development, a crucial step in overhauling many industries in the face of a major shortage in AI expertise.
For example, the DataRobot platform automatically cleans and reformats the raw data it receives, and then runs many competing algorithms against it, ranking their performance until arriving at an optimal approach.
In addition, a growing body of research focuses on self-learning systems, as Google is attempting to accomplish. Earlier this year the company demonstrated automated deep-learning neural network tuning.
"The goal is to make this technology more accessible," says Google researcher John Giannandrea.
Meanwhile, Carnegie Mellon University professor Eric Xing is developing an operating system built from different machine-learning elements, including virtualization, to abstract the complexity in designing and teaching AI.
Microsoft's Rich Caruana says there are many hazards in AI automation, warning against trusting too much in automated systems.
From Technology Review
View Full Article - May Require Free Registration
Abstracts Copyright © 2017 Information Inc., Bethesda, Maryland, USA
No entries found