#Google App Puts Neural networks on Your Phone to Translate Signs Offline In recent years Google has used networks of crudely simulated neurons running in its data centers to improve its speech recognition,
build software that learned to spot cats from Youtube videos, and power a photo storage service that knows what in your snaps.
Now the company wants you install artificial neural networks on your phone. Built into an updated version of Google translation app released today,
the technology expands its ability to translate printed text such as menus in a live view through your phone camera.
The app could previously translate between seven different languages. Now it can handle 27 and translate between them without an Internet connection.
That possible because Google engineers created slimmed-down versions of the artificial neural networks it uses in a technique called deep learning (see 0 Breakthrough Technologies 2013:
Deep Learning. They live inside the translation app and recognize the characters used by the different languages,
Google engineers first trained much larger and more powerful neural networks to find and recognize different letters.
This blog post has more details on how. It the first time Google has used that trick,
but it likely won be the last. Embedding the intelligence that artificial neural networks can provide into gadgets
so they don have to link to the Internet for tasks has clear benefits. Google is not the only company exploring that idea.
Coming changes to the design of the chips and software on mobile devices will make it easier and more powerful.
Mobile-chip maker Qualcomm has shown off a camera app with artificial neural networks inside that can recognize some objects
and More Other companies are also working on hardware that could run neural nets inside gadgets, robots,
and cars (see ilicon Chips That See Are Going to Make Your Smartphone Brilliant S
#Why the Time Seems Right for a Space-Based Internet Service providing Internet access from orbiting satellites concept that seemed to have died with the excesses of the dot-com boomas returned thanks to Spacex founder
(and dot-com billionaire) Elon musk. And while such a service would be expensive and risky to deploy,
000 microsatellites to provide broadband Internet services around the globe. Spacex is partnering with Google and Fidelity Investments,
which are investing $1 billion for a 10 percent stake in the endeavor. Richard Branson Virgin galactic and Qualcomm
and they can transmit huge amounts of data. Beaming data from a satellite is done by radio,
and is limited by the available spectrum, as well as the amount of power a satellite can get from its solar panels.
Most communications satellites have data-transfer speeds of around a gigabit per second, compared to several terabits per second for the fastest fiber.
But some things have changed since the late 1990s. For one thing, satellite technology has advanced, bringing the cost of deployment down significantly.
Spacex and Virgin galactic also hope to ride a different boom by targeting parts of the world where there is little infrastructure and a huge opportunity for Internet growth.
but Musk has stated that his Internet service would be aimed primarily at providing service to remote areas of the globe. oue got large swaths of land where there is a relatively low density of users,
Besides investing in Musk project, Google is working on a high-altitude balloon-based Internet delivery system called Loon.
And Facebook is developing high-altitude, high-endurance drones to deliver Internet capability to remote areas.
The Google and Facebook projects would be similar in concept to the space-based systems,
while operating within the Earth atmosphere. Whether, as Musk has suggested, Spacex service could also be a viable alternative for customers in the developed world is less certain.
#Google#s Modular Smartphone to Debut in Puerto rico Google vision of cheap modular smartphones made up of interchangeable pieces is getting closer to reality.
Google Advanced Technology and Projects (ATAP) group revealed the Spiral 2 on Wednesday during a crowded event for software developers at a Google office in Mountain view, California.
its face dominated by a large display and a receiver module for phone calls that includes a light and proximity sensor on the front.
and rectangular modules that snap into the phone slim metal skeleton to add different functions to the device.
The modules on the back include a camera, USB charger, Wi-fi, and Bluetooth radios, and the device main processor.
The display and receiver module on the front can also be swapped out. All the modules are held in place by magnets in the device frame.
Google says that this modular approach, which it calls Project Ara, will lead to inexpensive smartphones that can be customized
and upgraded by users depending on their budget or personal requirements. You might want a camera with an optical zoom lens for a sightseeing trip
for instance, a pollution sensor to keep tabs on asthma, or just a simple phone with basic capabilities.
But the Spiral 2 prototype also shows how tricky it will be for Google to make a modular gadget successful.
The latest device comes about nine months after ATAP unveiled the first Project Ara smartphone prototype, Spiral 1,
at a similar conference (see or Project Ara, It Moduleot Appdeas Wanted. That earlier device had Wi-fi but no working cellular connection,
and failed to work when presented to the crowd. Spiral 2 did manage to power up in public
and showed Google android operating system on its display. The Spiral 2 prototype is slimmer and looks more polished than the previous version.
But the handset is still far from being the flashy device with night vision and a slew of other modules imagined in a promotional Project Ara video shown at the conference.
Google prototypes have yet to catch up to existing smartphones on features such as cellular data speeds, for example.
Ara Knaian, the chief mechanical engineer for the project and also its namesake, said one tricky problem now solved is that the modules of the first prototype would disconnect themselves
Google says that device will be able to wirelessly transfer data between the modules and endoskeleton, eliminating the need for the spring-pin connectors used currently.
Project Ara leader, Paul Eremenko, said the group hopes to get the phone to last for a full day, at least,
though he then said that this might only happen by swapping in another battery at some point (Project Ara aims to let users do this without needing to turn off the phone).
The design of the phone skeleton will need also to be modified so that it can be manufactured easily.
Google plans to start selling Ara phones to customers in Puerto rico as a test market at some point this year, in collaboration with two wireless carriers.
The phones will be sold from trucks that function as mobile stores, as well as over the Internet. Just as third party apps were crucial to the popularity of smartphones,
getting people interested in modular phones will hinge on companies other than Google offering a diverse selection of modules.
Google has started already courting hardware companies to become module developers. On Wednesday the company showed a preview of the online ra Marketplacethat will let developers sell Google-approved modules to consumers.
Google is aiming to have a minimum of 20 modules available when it starts selling phones in Puerto rico. Google has also come up with 11 different reference designs for modules for developers to use.
Electronics giant Toshiba and chip makers Marvell and Nvidia are among those already showing interest;
Toshiba showed off some early camera modules at the conference. But the individual hardware developers I spoke with in the audience were still in the earliest stages of working on modules.
How much Project Ara modules might cost is unclear, but it must be low if Google is to attract people looking for their first smartphone.
Eremenko said in April last year that the bill of materials for a basic Ara handset complete with display, Wi-fi, battery,
and processor modules would run about $50, though he didn put an estimated sale price on it.
On Wednesday he declined to update that figure l
#CES 2015: Nvidia Demos a Car Computer Trained with Deep Learning Many cars now include cameras or other sensors that record the passing world and trigger intelligent behavior,
such as automatic braking or steering to avoid an obstacle. Today systems are usually unable to tell the difference between a trash can
and a traffic cop standing next to it, though. This week at the International Consumer electronics Show in Las vegas, Nvidia, a leading marking of computer graphics chips, unveiled a vehicle computer called the Drive PX that could help cars interpret
and react to the world around them. Nvidia already supplies chips to many car makers but engineers at those companies usually have to write software to collect
and process data from various different sensor systems. Drive PX is more powerful than existing hardware,
and it should also make it easier to integrate and process sensor data. The computer uses Nvidia new graphics microprocessor, the Tegra X1.
It is capable of processing information from up to 12 cameras simultaneously, and it comes with software designed to assist with safety or autonomous driving systems.
Most impressive, it includes a system trained to recognize different objects using a powerful technique known as deep learning (see 0 Breakthrough Technologies 2013:
Deep Learning. Another computer from Nvidia, called the Drive CX, is designed to generate realistic 3-D maps
and other graphics for dashboard displays. t pretty cool to bring this level of powerful computation into cars,
said John Leonard, a professor of mechanical engineering at MIT, who works on autonomous car technology. t the first such computer that seems really designed for a carn autopilot computer.
The new Nvidia hardware can also be updated remotely, so that car manufacturers can fix bugs or add new functionality.
This is something few car companies, aside from Tesla, do currently. So far Audi has emerged as an early buyer;
at CES, the company showed off a luxury concept car called the Audi Prologue that includes the Drive PX.
A year ago, the company announced at CES that it had developed a compact computer for processing sensor information (see udi Shows Off a Compact Brain for Self-driving cars.
That, too, included Nvidia chips. The introduction of Nvidia product is a landmark moment for deep learning,
a technology that processes sensory information efficiently by loosely mimicking the way the brain works.
At CES, Nvidia showed that its software can detect objects such as cars, people, bicycles and signs, even when they are hidden partly.
Yoshua Bengio, a deep-learning researcher at the University of Montreal, says the Nvidia chipset is an important commercial milestone. would not call it a breakthrough,
but more a continuous advance in a direction that has been going for a number of years now,
he said. Yann Lecun, a data scientist at New york University who leads deep-learning efforts at Facebook (see acebook Launches Advanced AI Effort to Find Meaning in Your Posts,
also sees the announcement as an important step: t is significant because current solutions tend to be closed
and proprietary, use custom and inflexible hardware, and tend to be lack boxesthat equipment manufacturers cannot really customize.
At a press event Sunday, Jen-Hsun Huang, Nvidia CEO, said the devices will provide ore computing horsepower inside a car than anything you have today. e
#Black Phosphorus: The Birth of a New Wonder Material In the last few years, two-dimensional crystals have emerged as some of the most exciting new materials to play with.
#AMD's new High Bandwidth Memory will power the graphics cards of the future Sitting down to talk graphics memory isn't usually anyone's idea of a good time
-and AMD just unveiled the memory technology it wants to power the GPUS of the future.
Seven years in the making, High Bandwidth Memory (HBM) promises to provide huge improvements over the GDDR5 memory currently sitting inside AMD's top-end graphics cards
It changes the layout of memory chips from the ground up-chips are stacked vertically like skyscraper floors in a 3d pattern (not unlike 3d technology used in SSD),
Graphics cards will soon be shrinking significantly thanks to HBM. A comparable 1gb graphics chip takes up just 6 percent of the surface area measured against GDDR5,
That could lead to top-end graphics cards half the size of today's, AMD says.
or the amount of data it can cope with at once. It means CPU and RAM performance can be increased without hitting a bottleneck on the GPU side of the equation."
"We needed a new memory that was far, far better from a power perspective,"said AMD's Joe Macri at a briefing with journalists."
"There's also a massive increase in bandwidth. We have blown basically through the bandwidth-per-watt wall...
"With Nvidia working on next-generation graphics technology of its own, battle between the two graphics giants will be renewed in earnest in the very near future-HBM is slated to appear in AMD's flagship graphics cards later this year.
AMD hasn't said whether that technology might be used or will be used in standard RAM; AMD sells a range of high-performance memory modules s
#Opinion: How Toshiba's new storage device could change the data centre Poor hard disk drive; advances in silicon manufacturing have allowed solid state drives to capture the headlines leaving the traditional spindle-spinning devices in the shadows.
But things are about to get more exciting for the latter. Toshiba made an important announcement yesterday as it unveiled a new solution that essentially is a new class of server (the Japanese company calls it a multi-device storage solution),
one that integrates, in an industry-standard, 3. 5-inch form factor, compute (64-bit,
probably ARM), networking (Gigabit Ethernet) and storage (with a pinch of SSD storage for low latency tasks and onboard RAM).
Future iterations we guess-could include an all-SSD model, a 2. 5-inch one or even, may be,
The whole set is enabled by an unidentified Linux platform that will allow the device to run what Toshiba calls, the next generation of software-defined storage applications.
The HDD becomes the server The implications for the industry are tremendous; such a solution could allow data centres to offer a richer set of scale-out object storage features.
From Toshiba's perspective, adding compute and networking features helps maintain margins while simplifying the overall calculation of total cost of ownership (TCO)
what it calls commodity servers used for storage management. The other announcement that dovetails nicely with it is that Toshiba has committed finally to deploying SMR technology, well after WD and Seagate.
Interestingly, the company managed to produce two products that shows its skills when it comes to cramming bits on a platter.
an external 2. 5-inch hard disk drive, managed to packs four 750gb platters in a tiny 15mm drive using perpendicular magnetic recording (PMR) only.
In addition, late last year, it debuted a 6tb hard disk drive that apparently squeezed six 1tb platters without the need of helium gas like HGST-or any other exotic technologies.
When we interviewed Seagate's Joe Fagan in March, he hinted that his company may
and more processing power to the hard drive as its remit goes beyond just holding bits.
if WD (or rather its subsidiary HGST) and Seagate do the same as Toshiba's move seems to be motivated by customer demand.
The algorithms allow the robot to build its knowledge slowly over time like humans do,
"The exact same software, which encodes how the robot can learn, was used to allow the robot to learn all the different tasks we gave it."
the robot was rewarded"with points allocated by the algorithm.""We still have a long way to go before our robots can learn to clean a house
#Processors do grow on trees: your next phone could be made of wood Engineers hunting for a way to make electronics more sustainable have hit on a novel invention-a semiconductor chip made almost entirely out of wood.
The idea is that instead of making chips from petroleum-based plastic, we'd be able to use cellulose nanofibril-a flexible,
and Canadian researchers has managed to successfully sequence the full genome of a living organism using a machine the size of a smartphone called the Minion.
Its tiny size and relatively low cost could allow scientists to perform much more advanced analysis away from a lab."The amazing thing about this device is that it is many times smaller than a normal sequencer-you just attach it to a laptop using a USB cable"
The drawbacks identified so far revolve around accuracy-the data it produces is currently less accurate than the data produced by a larger machine,
but it is producing a lot of data, "said Gregory Buck, also from the Center from the Study of Biological Complexity."
#Google maps can make sure you never see a"Closed"sign again Google maps has got another incredibly useful feature thanks to a recent update that will warn you
Instead, when you set off Google maps will now show a message that reads"Your destination may be closed
At the moment the update is only available for the Android version of Google maps, but we've reached out to Google to find out
when it will be coming to ios. Via The Telegrap p
#Scientists are turning your gut cells into computers A team of biological engineers at the Massachusetts institute of technology are turning the cells in our guts into computers.
It's hoped that this could one day allow us to program those cells to detect
and treat diseases. They've published the details of a series of sensors, memory switches and circuits that can be encoded into the human gut bacterium Bacteroides thetaiotaomicron.
Eventually, it's hoped that the resulting gut computers could help the early detection and treatment of disorders like inflammatory bowel disease and colon cancer r
French President, Francois Hollande, said on Tuesday in Paris the government will present a draft law next month that makes Internet operators ccomplicesof hate-speech offenses
Interior Minister Bernard Cazeneuve said he will travel to the U s. to seek help from the heads of Twitter, Microsoft, Google, and Facebook.
This is not just leading to the oppression of every internet user but also makes little sense as potential terrorists will move away from public domains controlled by such large companies to darker areas of the web where they are monitored less-easily.
Moving away from terrorism, surveillance is a controversial subject on a global scale thanks to whistleblowers such as Edward Snowden
BT and EE partnership will reduce'not spots'Broadband provider BT has confirmed its intention to acquire the UK's biggest mobile network
which could be a major step in providing coverage to mobile not spots where residents are struggling with limited or complete lack of connectivity.
Earlier this week the UK government published a map of public digital infrastructure (PDF) in the hope of tackling some of the mobile'not spots'across the country.
Minister of state for culture and the digital economy Ed Vaizey said: This report will be instrumental in driving savings
The telecoms giant has over seven million broadband subscribers many of which will use the company's own'Home Hub'routers.
Although new hardware will be required if BT chooses it could roll out new Home Hub devices which feature femtocell technology to boost mobile coverage in rural areas for its EE purchase.
A femtocell is a small low-power cellular base station which is designed typically for use in a home or small business to boost coverage.
The firm which was created out of a partnership between T-Mobile and Orange had promised to connect more than 1500 rural communities within three years by investing in small cells to extend coverage.
which believes that Wi-fi offload will dominate traffic by 2017 to deal with capacity demands from a mobile data traffic increase of more than sevenfold between 2014 and 2019 as part of a wider global growth of almost ten times.
The ongoing adoption of more powerful mobile devices and wider deployments of emerging M2m applications combined with broader access to faster wireless networks will be key contributors to significant mobile traffic growth in the coming years.
This mobile-centric environment will give service providers a new landscape of challenges and opportunities to innovatively deliver a variety of mobile services
and experiences to consumers and business users as the Internet of Everything (Ioe) continues to take shape.
#Human Genes Can Save Yeast WIKIPEDIA, LILLY MA large number of human genes can substitute for their defective counterparts in yeast
Incorporating data from previously reported substitution experiments, that percentage rose to 47. The team went on to examine
told The Scientist in an e-mail. ny technological advancement that can add in the precision and control of genetic modification is an important advance,
The idea is that users learn how to interpret the tingling patterns and come to eetheir environment through the gadget.
Users wear glasses mounted with a video camera and suck on an electrode array about the size of a lollipop.
and an early user of the Brainport, told the Washington post back in 2009. here is not that picture in your head.
but it a first step. or five years I have stared at a blank, black screen, Jernigan told the Post. eople are thinking outside of the box,
Mesulam told Motherboard o
#Diagnosing Ebola in 15 Minutes As West Africa has battled Ebola over the last year, clinicians have been restricted to time-consuming,
In the first, the researchers used electrodes to link the brains of three monkeys to a computer
and allowed each animal to contribute their thoughts to controlling an image of a robotic arm on a screen in front of them.
In a second study, on rats, the team linked the rodentsbrains not just to a computer
The researchers then showed the animals images of a robotic arm on the screen and gave each monkey control over certain parts of its movementither a single axis (x or y) or two dimensions (x-y, y-z,
Perhaps such a brainet could help an experienced user train someone with a new limb to control it,
#Semiconductor crystals could be key to extending Moore Law IBM researchers have developed a process for growing crystals made from semiconductor materials,
which they claim can then be integrated onto silicon chips in a crucial step forward for the future of computing.
an observation made by Intel cofounder Graham Moore in 1965 that the number of transistors per square inch on integrated circuits doubled every year.
a researcher with IBM Research Gmbh at Zurich Research Laboratory in Switzerland and the lead author of the paper. e need better performing transistors as we continue down-scaling,
III-V semiconductors are believed to be a potential future material for computer chips but their integration onto silicon has been unsuccessful up until now.
but the details of the dynamics are difficult to monitor, said Ahmad Kabbani, a professor of chemistry at the Lebanese American University,
Arranging these artificial chromatophores in a linear array, the team devised mathematical algorithms to control how the cells react to changes in state of neighbouring cells,
and devise new algorithms to simulate patterns in two-dimensional chromophore arrays o
#Zero emissions liquid air engine begins tests 26 june 2015uk liquid-nitrogen engine pioneer Dearman has begun full testing of its zero-emission engine technology at its new liquid air R&d facility.
he said. his is building on work we have done using voice control on consoles like the Xbox One, for example,
the BBC developed a prototype mind control TV using a low-cost headset equipped with sensors that measure electrical activity in the brain.
The electroencephalography (EEG) brainwave reading headset has a sensor that rests on the user forehead,
The user can choose to operate the device in either oncentrationor editationmode. If they choose meditation
the headset and app monitor their level of relaxation, which is displayed on a volume bar on the side of the screen. hen,
when a certain threshold is reached for that type of electrical activity, it sends a signal to the device on our tablet
which in turn sends a signal to the TV, said Saihan. During the experiment, 10 users were given a headset to wear,
and sat in front of the TV. The users either concentrated hard or relaxed their brain until the volume bar showed the threshold had been reached, at
which point a signal was sent to the TV to open the application, an experimental form of iplayer.
The users were presented then with a screen showing the five most popular programmes on iplayer at that time.
To select a programme the users waited until the show was highlighted, and then relaxed until the volume bar again reached the necessary threshold.
Some users taking part in the experiment found the technology easier to use than others.
and currently only allows users to select either nor ff said Saihan. Whether the idea takes off will ultimately depend on how the technology evolves over the coming years,
technologists and other users an idea of how the technology might be used in the future
#Cold plasma technique helps wounds heal quicker 29 june 2015medical researchers in Germany have developed a technique which,
about the size of a hand-held torch, houses an electrode that the doctor holds close to the wound site.
The result is only about the size of a laptop and can be plugged into a normal socket between 100 and 230 V,
Overtext Web Module V3.0 Alpha
Copyright Semantic-Knowledge, 1994-2011