Enterprise Mobile News of the Week: 2/2/18

February is off to a great start. Lots of interesting enterprise mobile news the last few weeks: a bunch of advances in machine learning from Google, some ways to trick Speech Recognition ML models, and a very interesting collaboration to (hopefully) reduce healthcare insurance and delivery costs by some parties you might not have expected.

• It’s one thing to have access to pre-trained generic ML models, it’s another to have the expertise to create models that match the specific needs of your company. Google is trying to make it a heck of a lot easier to build custom ML models with its new Cloud AutoML service. Their first installment, AutoML Vision, abstracts away most of the complexity, using a “drag-and-drop interface [that] lets you easily upload images, train and manage models, and then deploy those trained models directly on Google Cloud.” Easy-peasy.
• While adoption is still in its very early stages, many organizations have recognized that Machine Learning will become a critical technology pillar to creating competitive advantage. Yet few recognize that scaling Machine Learning in the enterprise requires (to borrow a phrase from the movie Taken) “a very particular set of skills.” Enter MLOps.
• Alphabet launches Chronicle, a new business unit focused on using Alphabet/Google’s expertise in machine learning to more rapidly and reliably detect cyber-security threats.
• While we’re waiting on the Trump Administration to privatize 5G infrastructure, here’s a story of interest to those developing apps for industrial settings and IoT deployments. MulteFire’s small cell deployments provide organizations bandwidth and secure connections via Private LTE networks using unlicensed spectrum where limitations or lack of traditional cell networks hamper the ability to achieve full potential from mobile and IoT initiatives. The MulteFire Alliance previously published the spec, but Nokia is the first to ship.
• So maybe there’s a little more work to be done when it comes to machine learning-driven speech recognition. Regardless, here’s a very interesting paper from some Cornell University researchers.
  

  Given any audio waveform, we can produce another that is over 99.9% similar, but transcribes as any phrase we choose (at a rate of up to 50 characters per second). We apply our iterative optimization-based attack to Mozilla’s implementation DeepSpeech end-to-end, and show it has a 100% success rate.”

• Huzzah! IT can finally block iOS updates with 11.3 for devices under Supervision Mode. One of great attributes of iOS has been the ability to easily update devices to the latest and greatest version, but this is often the bane of developers and admins who still want to test the impact of new OS releases on the apps they’ve developed and deployed. So having more time to test and remediate any issues is sure to get a warm welcome.
• Nothing seems to increase faster than health insurance costs, and this doesn’t even factor in the effects of removing the individual mandate to smooth out the risk pools. Good news is there may be some relief in sight as Amazon, Berkshire-Hathaway and JPMorganChase try to tackle reducing healthcare costs while increasing quality of care with technology and data. With Bezos and Buffet involved, color me excited.
• Finally, our great national nightmare is almost over—iOS 11.3 is going to solve the calculator problem.