Insight. Applied.
Google has just release performance data for its Tensor Processing Unit (TPU) custom machine learning chips.
TPUs are
Google claim that without these chips they would have to double the number of data centers they run.
Matt Turck has published the 2017 edition of his annual review of the big data landscape. It include an infographic showing the key players in various sub-fields.
This year’s main finding was that
Big Data provides the pipes, and AI provides the smarts.
Leo Laporte and guests discuss social media profiling in the latest TWiT podcast.
They describe credit ratings in China that are formed not just from financial data, but also social networking activities.
According to the podcast these scores have already been used deny train travel to 7m people.
As Laporte says, “It’s out of Black Mirror.”
I thought the CIA’s “best practices” for coding malware were interesting.
It’s basically a list of things you have to consider when dealing with a sophisticated attacker. Basic audit features aren’t enough if your data and systems present an attractive target.
Over half a million UK National Health Service (NHS) records went missing between 2011 and 2016.
When I first skimmed the headline, I was thinking “Another data breach…[yawn].” But, turns out it was all paper. 500,000 paper records went missing.
A private mail direction company sent the records to a warehouse. They had either been incorrectly addressed or the associated patients had moved surgery. Unfortunately, the records—including test results for serious illnesses—were abandoned at that point.
A review is underway to decide whether any patients died due to the missing documents.
You can see how databases are overlooked—the information is pretty abstract. But, a mountain of 708,000 documents (there were administrative documents as well)?! Who walked by that every day without asking questions?