Wikiverse turns Wikipedia into a visual universe of articles

Source: https://www.engadget.com/2016/08/18/wikiverse-turns-wikipedia-into-a-visual-universe-of-articles/

I've spent (too) many lazy evenings hopping from one Wikipedia entry to the next, reading things that could be useful for trivia night. It's fun enough, but this website that visualizes Wiki as a universe could make the experience even better for a lot of people. You can zoom around to visit clusters of stars representing interconnected topics -- clicking on one will load the article itself right within the interface. Since each star is visually connected to related entries with colored loopy lines, you can hop around like you would on the actual Wikipedia website.

Wikiverse is the upgraded version of an old Chrome experiment we covered back in 2014 called WikiGalaxy. It's designed by French computer scientist Owen Cornec who wanted to find a way to display entries in "a more engaging way." From what I can see, the new version is a lot more colorful compared to the older one. Cornec originally wanted to color code stars according to their super-categories, but he told us he wasn't able to do so because there are too many super-categories to count. Instead, he decided to use colors to differentiate clusters from each other and to indicate whether an entry is within that cluster or in another one. Wikiverse also feels faster than its predecessor, even on browsers other than Chrome, so you'll have zero issues pretending to be a spacecraft exploring a universe of knowledge.

Source: Wikiverse

Read More...

Summary only...

Samsung plugs IBM's brain-imitating chip into an advanced sensor

Source: https://www.engadget.com/2016/08/14/samsung-ibm-truenorth-chip-advanced-sensor/

IBM's TrueNorth, a so-called "cognitive chip," remarkably resembles the human brain: its 4,096 cores combine to create about a million digital neurons and 256 million synapse connections. In short, like everyone's favorite complex organ, it operates extremely quickly and consumes far less energy than typical processors. Samsung has taken the chip and plugged it into its Dynamic Vision Sensor (DVS) to process digital imagery at a blindingly fast rate.

Typical digital cameras max out 120 frames per second, but a DVS-equipped gadget can capture an incredible 2,000 fps. Unlike a conventional sensor, each pixel on Samsung's only reacts if it needs to report a change in what it's seeing, according to CNET. That high speed could be useful for creating 3D maps or gesture controls. At a press event on Thursday in San Jose, the company demonstrated its ability to control a TV as it recognized hand waves and finger pinches from ten feet away.

DVS is efficient like its TrueNorth chip base, and only consumes about 300 milliwatts of power. That's about a hundredth the drain of a laptop's processor and a tenth of a phone's, a Samsung VP said at the event. But we still have a ways to go before we approach the minimal power requirements of the human brain, he said, which can process some tasks at 100 million times less power than a computer.

Source: CNET

Read More...

Summary only...

Intel buys an AI processing powerhouse

Source: https://www.engadget.com/2016/08/09/intel-acquires-nervana/

Intel isn't the biggest player in the artificial intelligence world, but it might change that very shortly. The chip giant just bought Nervana, a startup specializing in processor-based "deep learning as a service" through technology like its upcoming Nervana Engine, a dedicated AI hardware accelerator due in early 2017. The Intel crew isn't shy about its intentions. It wants Nervana's expertise to help boost the AI performance of its Xeon and Xeon Phi processors on top of broadening its AI know-how -- it's a fast track to the future.

The buyout may well be a hedge against NVIDIA. The graphics firm has been big on accelerating AI in both self-driving car systems and its high-powered graphics cards. It could face serious challenges if Intel offers similar power, especially if it's in a run-of-the-mill server that doesn't need pricey extras to handle machine learning tasks. Whether or not Intel is targeting NVIDIA, there's no question that it's getting serious about one of the hottest fields in tech.

Source: Intel, Playground

Read More...

Summary only...

Change up your space with robotic furniture from Ori

Source: https://www.engadget.com/2016/07/11/ori-robotic-furniture/

Need a way to dramatically improve your living space? How about robotic furniture? No, it's not some far-off dream for the future. It's a collaboration between MIT Media Lab spin-off Ori with designer Yves Béhar.

Ori is introducing a line of architectural robotics, as they're described, which are meant to transform the way we arrange and decorate our bedrooms. Using these special pieces of furniture, you could theoretically take a studio apartment and transform it into a much larger space with plenty more functionality.

The line of connected products is set to offer innovation for those in urban areas, where simply pressing a button can change a living room into a bedroom, walk-in closet or even an office connected via "smart-home ecosystem." In fact, the Ori line gets its name from "origami," which makes sense when you consider The first residential versions of the furniture will be rolled out by developers in Boston, Washington DC and Seattle this summer and the program will be accepting pilot program partners in early 2017.

Read More...

Summary only...

Bitcoin mining just became more difficult, on purpose

Source: https://www.engadget.com/2016/07/09/bitcoin-mining-value-cut-in-half/

If you have computers chugging away as bitcoin mining machines, don't be surprised if your output just fell through the floor. Reuters notes that code built into the digital currency system has cut the mining reward in half as of July 9th. Where there were previously 25 bitcoins (roughly $16,000) to be mined every 10 minutes, you now have to fight over 12.5. The measure automatically kicks in every four years as part of an attempt to curb inflation that would come from both a growing number of miners and ever-faster computers.

To no one's surprise, reducing the reward could have serious consequences for dedicated miners. As you have to work twice as hard to get the same money, companies with not-so-efficient operations may have no choice but to restructure or even close shop entirely. KnCMiner, for instance, declared bankruptcy in May after warning about the impending profit loss. Those miners most likely to survive are the ones that keep costs to a minimum through lower-power computers and minimal staff.

You'd think that the industry would have anticipated the halving given that it will happen every four years like clockwork, but that's not necessarily the case. Although bitcoin isn't quite as celebrated as it was a while back, it's still far more mainstream than it was in 2012. There are many more people mining than there were four years ago, and not all of them realize that they'll have to factor in those reward cuts. The bigger question is whether or not the bitcoin business will be better-prepared when 2020 rolls around. Miners will either have to trim costs yet again or hope that they can make money from transaction fees.

Source: Reuters

Read More...

Summary only...