I came across a lot of interesting items today spanning cloud, artificial intelligence and big data, so please do read the links below so you don’t miss them. However, these are the three that, for a variety of reasons, really stuck out:
IBM launches new version of its private cloud with built-in Kubernetes support (GeekWire):
This is IBM’s latest attempt to deliver hybrid clouds, this time using Docker and Kubernetes as the connective tissue between public cloud and on-prem environments. What’s underappreciated, I think, is just how many WebSphere apps, DB2 deployments and the like are floating around among IBM customers, all of which the company is promising to containerize with this new offering. If IBM has a hook to keep customers on its cloud as they look to modernize, perhaps this is it. If you check out the podcast tomorrow, you can hear more about this from IBM Fellow Bala Rajaraman.
Google’s AI wizard unveils a new twist on neural networks (WIRED):
Geoff Hinton’s research into “capsule networks” has been written about before, but now there are research papers to go along with his theory that traditional neural-network-based approaches to image recognition are lacking. Google has a lot of irons in the AI fire at the moment, a few of which have the potential to be game-changers if they can prove themselves beyond the lab. And, notably, as work from DeepMind, Hinton and others strays further away from traditional deep learning techniques, Google could find itself that much further ahead of the pack once again.
Crowdsourcing big-data analysis (MIT):
There is still so much research, and even commercial, effort going into data science collaboration that it must not be a solved problem – even despite, well, all the effort that has been put into it. However, I think this research from MIT, called FeatureHub, is cool because it might actually prove useful on problems that are too big for a normal-sized team and possibly tough to monetize. As one of the researchers put it:
“I think that the concept of massive and open data science can be really leveraged for areas where there’s a strong social impact but not necessarily a single profit-making or government organization that is coordinating responses.“