Your weekly tech round up:
- Google starts experimenting with new shapes for hard drives
- Facebook’s AI Algorithm Maps the World
- Searching for the Right Image
- Needing more wireless infrastructure, Facebook will design its own
- Hush Money for Hackers
Google Starts Experimenting with New Shapes for Hard Drives
Some things have changed in hard disk drives over the years, like density and reliability, but one thing that’s remained the same is its size (aside from a few thickness tweaks along the way). Well for Google, upping storage density is no longer the key to storage, so they’ve decided to take matters into their own hands. When you own the likes of YouTube, which requires one petabyte of new storage every single day, Google is experimenting with hard drives that are taller, or developed as clusters to delivers performance at lowers costs.
Analyzing 14.6 Billion Images – Facebook’s AI Maps the World
Facebook is focused on bringing internet connectivity to rural and remote communities by flying internet beaming drones, satellites and laser technology. But in order to do that, they need to know where people are actually living and how.
Using 350 terabytes of maps from Columbia University and the World Bank Facebook employed artificial intelligence (AI) algorithms to process 14.6 billion images (running on thousands of servers simultaneously) to identify human-built structures. The result? The most detailed map of human settlement in 20 countries, down to 5 meters of accuracy, the typical range of a Wi-Fi connection.
The machine-learning algorithms can now map more of Earth in a week than we have in history. Facebook believes the maps could be put to other good uses and will make them publicly available later in the year – read more on the future of human connectivity and AI here.
Searching for the Right Image
Winston Hsu at National Taiwan University has been behind some of the more interesting developments in image search over the last several years. He (and fellow researchers and graduate students) developed a search app called Flora that could identify flowers in photographs. They’re also working on a project to find pictures of individuals from a sketch—you can imagine the applications in police work.
Searching unstructured data like images requires massive computing power and even larger amounts of data. Think about all of the spatial parameters a single picture contains: it’s a classic Big Data problem. Nonetheless, as Hsu’s work shows, we’re getting closer and that means broader commercialization for ideas like this is probably on the horizon.
EE Times has more.
Needing More Wireless Infrastructure, Facebook will Design its Own
If you build it, they will come. If they’re taking too long, do it yourself. That’s what Facebook is doing with their Telecom Infra Project (TIP), leveraging open source hardware and software to accelerate the development of telecommunications networking. Data is scaling faster than the current telecom infrastructure can keep up, so if the likes of Facebook and Google want more users on their network, they need to forge ahead and make 5G connectivity happen fast. Read more on the project here.
Hush Money for Hackers
Computerworld reports that certain companies have paid up to $1 million to hackers to keep them quiet about security breaches. Hackers break into the network, review the data and then try to extort the original owners. It’s a disturbing trend to say the least. “We have seen seven-figure payouts by organizations that are afraid for that data to be published,” said Charles Carmakal, a vice president with Mandiant, a division of FireEye.
But it also points up another problem: people aren’t using encryption. Encryption can’t stop attacks, but it can dramatically reduce the value of data. The ability to encrypt data is also becoming far easier because of self-encrypting technology embedded in many SSDs.
Unfortunately, as these cyber blackmail incidents show, the message hasn’t gotten through to everyone yet.