Facebook is using AI to make language translation nine times faster

Why it matters to you

If you’re ever had dealings with someone who speaks a foreign language, you probably know how difficult a language barrier can be. Facebook’s new tech could change that.

Artificially intelligent systems are only getting better, and they’re likely to appear in our computers and on our phones more and more often over the next few years. Facebook has been using artificial intelligence and machine learning for various things — like its M digital assistant — but now the company is turning to AI for another purpose: translation.

Facebook’s research team has published a report on how artificial intelligence is a hefty nine times faster than traditional language translation software. Not only that, but the researchers have revealed that the source code for the translation software is open-source, so anyone can get their hands on it to verify the results.

The report highlights the use of convolutional neural networks (CNN) as opposed to recurrent neural networks (RNN), which translate sentences one word at a time in a linear order. The new architecture, however, can take words further down in the sentence into consideration during the translation process, which helps make the translation far more accurate. This actually marks the first time a CNN has managed to outperform an RNN in language translation, and Facebook now hopes to expand it to to cover more languages.

“Language translation is important to Facebook’s mission of making the world more open and connected, enabling everyone to consume posts or videos in their preferred language — all at the highest possible accuracy and speed,” said the company in a blog post.

Convolutional Neural Networks aren’t a totally new technology, but they haven’t really been applied to translation before. As a result of the new tech, Facebook can compute different aspects of a sentence at the same time, and as a result it can train its systems using a lot less computational power — which in turn results in faster translation. The system is also open source, meaning translation should get better across the web — not just in Facebook’s offerings.

Facebook is using AI to make language translation nine times faster

Why it matters to you

If you’re ever had dealings with someone who speaks a foreign language, you probably know how difficult a language barrier can be. Facebook’s new tech could change that.

Artificially intelligent systems are only getting better, and they’re likely to appear in our computers and on our phones more and more often over the next few years. Facebook has been using artificial intelligence and machine learning for various things — like its M digital assistant — but now the company is turning to AI for another purpose: translation.

Facebook’s research team has published a report on how artificial intelligence is a hefty nine times faster than traditional language translation software. Not only that, but the researchers have revealed that the source code for the translation software is open-source, so anyone can get their hands on it to verify the results.

The report highlights the use of convolutional neural networks (CNN) as opposed to recurrent neural networks (RNN), which translate sentences one word at a time in a linear order. The new architecture, however, can take words further down in the sentence into consideration during the translation process, which helps make the translation far more accurate. This actually marks the first time a CNN has managed to outperform an RNN in language translation, and Facebook now hopes to expand it to to cover more languages.

“Language translation is important to Facebook’s mission of making the world more open and connected, enabling everyone to consume posts or videos in their preferred language — all at the highest possible accuracy and speed,” said the company in a blog post.

Convolutional Neural Networks aren’t a totally new technology, but they haven’t really been applied to translation before. As a result of the new tech, Facebook can compute different aspects of a sentence at the same time, and as a result it can train its systems using a lot less computational power — which in turn results in faster translation. The system is also open source, meaning translation should get better across the web — not just in Facebook’s offerings.

Microsoft invites Insiders to discover what’s new in OneNote for Windows 10

Why it matters to you

You can enter for a chance to win prizes by tweeting new features in the May 2017 version of OneNote for Windows 10 — but only if you’re an Office Insider.

Microsoft introduced a new version of OneNote for Windows 10 on Tuesday, and it is available exclusively for Office Insiders who are running on the Fast Ring and have updated to the May Insiders version of OneNote. The company is doing things differently this time around — rather than publishing a changelog for the new version, it is inviting users on a quest to find the new features themselves.

#OfficeInsiders It's time–Grab your morning brew & get ready for the festivities! https://t.co/KWWJa9aFhb. #OneNoteQuest #TwitterTakeover pic.twitter.com/6sHbZ8oavJ

— MS Office Insiders (@OfficeInsider) May 9, 2017

If you head over to the Microsoft Office Insider Twitter account, you can follow along using the hashtag #OneNoteQuest. That’s where you can pose your questions to the Office Insider team, identify features that you have discovered, and, perhaps most important, have a chance to win prizes.

Mark Coppock/Digital Trends

In order to participate, you will need to install the latest Office Insider Fast Ring update for OneNote for Windows 10 from the Windows Store. Head to OneNote and open Settings, then Options, and then turn on Enable Beta Features. Finally, scan the app for new features and tweet them to @OfficeInsider with the feature name and the OneNote build you are currently running.

Digging through the various tweets, here are some of the new features that users have discovered:

• You can now open multiple windows in OneNote for Windows 10, making it easier to access and manage your notes.
• You can choose which navigation panes to keep open, including always showing the notebook pane.
• You can nickname notebooks and change notebook colors.
• OneNote for Windows 10 will now show you which pages have unread changes, and you can mark them as read.
• You can insert meeting details in OneNote for Windows 10.
• The Send to OneNote app, which you can install from the Window Store, can be used to send to OneNote for Windows 10 from any app via the print dialogue.

Mark Coppock/Digital Trends

Even if you miss the festivities, you can keep scanning the tweets for new features. OneNote for Windows 10 is a nice and lightweight alternative to the desktop OneNote 2016 app, and these new features bring some welcome functionality. Of course, you will need to be an Office Insider to participate, and you can join up for the program here.

Microsoft Build 2017 is happening! Here’s how to follow it

Catch all the Microsoft Build announcements over at Windows Central!

Microsoft Build is like Google I/O, but rainier. I kid, I kid. This year, Microsoft’s annual developer takes place in the company’s hometown of Seattle, Washed Out (don’t @ me, Andrew).

Starting May 10 at 8am PT / 11am ET, the event kicks off with a keynote by CEO Satya Nadella, and even though Microsoft just announced its new Chromebook competitor, Windows 10 S, along with a pretty new laptop, the Surface Laptop, there will certainly be some big announcements across the company’s vast product line — including Cortana and some other Android-based initiatives.

On day two, Terry Myerson, Executive Vice President of the Windows and Devices Group, gets on stage at 8:30am PT / 11:30am ET to talk about what’s happening with Windows 10 and, perhaps, its smartphone strategy.

Microsoft Build Day 1 + 2 Keynote liveblog and livestream

Our buddies at Windows Central will be covering everything, including all the announcements, analysis and lack of smartphones. I kid, I kid. Sort of.

How VW’s Arteon keeps you alive if you pass out while driving

The Volkswagen Arteon is not a fully fledged self-driving car, but it definitely has some fancy autonomous features. The sedan, which is due to land in the United States this summer, features a program that will recognize if a driver has blacked out at the wheel and then gently steer the car to the side of the road.

The system is called Emergency Assist 2.0, and it takes advantage of four existing VW safety programs: Adaptive Cruise Control, Side Assist, Lane Assist and Park Assist. The Arteon will notice if a driver hasn’t touched the gas, brakes or steering wheel for a certain length of time, and it will attempt to “wake up” the person with sounds, visual cues and even a physical brake tap. If that doesn’t work, Emergency Assist 2.0 takes over: The car’s hazard lights switch on and it steers itself to a safe, nearside lane.

The system uses Lane Assist to recognize lines on the road and Park Assist to actually steer the car. Radar detection via Adaptive Cruise Control and Side Assist helps ensure the Arteon won’t hit any other vehicles.

This type of technology isn’t new. Vehicle manufacturers like BMW have been toying with similar emergency systems since 2009, at least. However, that was back when driverless cars were still the stuff of pipe dreams. Today, autonomous-vehicle technology is becoming more ubiquitous by the second, and the Arteon’s Emergency Assist 2.0 — a self-driving system in a traditionally controlled car — is a testament to this growing industry.

President Trump fires FBI Director James Comey

President Donald Trump has fired FBI Director James Comey, who recently led multiple investigations into Hillary Clinton’s use of a private email server when she was Secretary of State and Russia’s involvement in the 2016 US presidential election. Trump acted on the recommendation of Attorney General Jeff Sessions and Deputy Attorney General Rod Rosenstein, who argued Comey caused “substantial damage” to the FBI’s reputation and credibility over the past year.

Which brings us back to Clinton’s emails.

“I cannot defend the Director’s handling of the conclusion of the investigation of Secretary Clinton’s emails, and I do not understand his refusal to accept the nearly universal judgment that he was mistaken,” Rosenstein writes.

BREAKING: FBI Director James Comey has been fired, White House says. https://t.co/8UupHPDBNs pic.twitter.com/U3Mxx8LY2p

— ABC News (@ABC) May 9, 2017

He’s talking specifically about Comey’s press conference on July 5th, 2016, when he announced the FBI would not bring charges against Clinton for her use of a private email server while she was Secretary of State, even though he said her actions were “extremely careless.” Rosenstein argues — and Trump and Sessions clearly agree — that Comey should not have closed the case without consulting members of the Justice Department.

“Compounding the error, the Director ignored another longstanding principle: we do not hold press conferences to release derogatory information about the subject of a declined criminal investigation. … The Director laid out his version of the facts for the news media as if it were a closing argument, but without a trial,” Rosenstein writes.

Just last week, Comey defended his handling of the Clinton email probe to the Senate Judiciary Committee. However, he mistakenly said Clinton aide Huma Abedin “forwarded hundreds and thousands of emails, some of which contain classified information,” to her husband at the time, former Congressman Anthony Weiner. The FBI later sent a letter to the committee saying this information wasn’t accurate.

Comey was also heading an investigation into whether members of the Trump campaign coordinated with Russia to influence the 2016 US presidential election. In his termination letter to Comey, Trump mentions this investigation and the fact that he has not been personally implicated in it, saying, “While I greatly appreciate you informing me, on three separate occasions, that I am not under investigation, I nevertheless concur with the judgment of the Department of Justice that you are not able to effectively lead the Bureau.”

Comey consistently supported decryption legislation, and argued for “backdoors” that would enable law enforcement to search locked phones and other devices. In March, Comey refuted Trump’s claim that former President Barack Obama wiretapped Trump Tower during the campaign.

Trump and Comey have publicly locked horns over the past year. On May 2nd, Trump tweeted the following message: “FBI Director Comey was the best thing that ever happened to Hillary Clinton in that he gave her a free pass for many bad deeds! The phony….”

Trump fired Comey one week after that tweet, as he was in Los Angeles, California, for work.

Comey learned of firing as he addressed FBI employees in LA. TV screens in background flashed news of firing. Letter was then given to FBI.

— Michael S. Schmidt (@nytmike) May 9, 2017

“The FBI is one of our Nation’s most cherished and respected institutions and today will mark a new beginning for our crown jewel of law enforcement,” Trump said today in a prepared statement.

Robot that performs surgery inside your eye passes clinical trial

The next time you go under the knife for retinal surgery, it may not be a human hand holding the blade. That’s because a revolutionary surgical system developed University of Oxford in the United Kingdom, which just passed its first set of clinical trials, is able to perform these intricate operations better than even the steadiest surgeon.

The problem lies in the pulse. Retinal surgeries rely on creating miniscule holes in the eye to gain access to the retina itself, a 10 micron thick flap of membrane that converts light into electrical signals that the brain can interpret. In the case of issues like an epiretinal membrane, essentially a scar on the retina caused by anything from injury to disease to just growing old, even the flow of blood through a surgeon’s hands is enough to throw off their accuracy, raising the odds that they’ll cut too deeply and make matters worse.

However Oxford’s Robotic Retinal Dissection Device (yes, “R2D2”) has no pulse and is therefore precise enough to move just 10 microns at a time, a tenfold increase over what human surgeons can achieve. That’s important when you’re shoving a tiny knife into a millimeter-wide hold in the side of the patient’s eyeball. The entire process is really still performed by a surgeon, albeit they’re holding a joystick rather than the scalpel itself.

The Oxford team has been developing the R2D2 since 2011 and conducted the first successful experimental procedure on a 70-year-old priest last September. Since then, the team has performed another 11 successful surgeries in a randomized clinical trial pitting it against results performed by humans.

These operations were not without their complications however. According to the team, who presented their findings at the the annual meeting of the Association for Research in Vision and Ophthalmology (ARVO) in Baltimore on Monday, two patients suffered minor bleeding (aka micro-hommorhages) and another suffered a “retinal touch” which heightens the risk that the retina will tear or detach. Still, that’s better than the control group which had five hemorrhages and a pair of retinal touches.

This isn’t the first time that robots have taken the helm in delicate eye surgeries. In January of this year, doctors at University Hospitals Leuven in Belgium successfully trialed a robotic syringe for treating retinal vein occlusion, wherein a vein in the retina becomes blocked and can lead to blindness. Treating that condition is insanely expensive, around $35,000 an eye, that requires surgeons to inject medicine into a 0.1mm-wide vein in the back of the eye for ten minutes straight.

The Oxford team believes that its R2D2 can do even more for patients. They hope to next outfit the machine with needles that can be used to inject fluids into the back of the eye, such as those used for retinal gene therapy.

Source: NBC News

Alibaba opens its mobile payment system to 4 million US stores

Chinese internet titan Alibaba has struck a deal to let its global AliPay customers buy things in the US through First Data, makers of the Clover point-of-sale system. Just don’t mistake this for a full American expansion to compete with Apple Pay and Android Pay: The partnership is explicitly aimed to convenience AliPay’s 450 million existing global users visiting the US.

Which isn’t to say that won’t happen in the future. This could be a move to get AliPay’s foot in the door to later open its service up to US consumers. For now, the deal allows current customers to make purchases through the service at four million First Data-connected merchants, saving them the fuss and expense of currency exchange or international credit card fees.

As The Verge points out, if you’re keen on keeping track of the global payment system scorecard, AliPay is now accepted in 70 countries compared to Apple Pay’s 15 or Android Pay’s 10. PayPal is still the leader by a longshot: Its service is supported in 200 countries and 25 currencies.

Via: TechCrunch

Source: First Data

Apple Acquires ‘Beddit’ iPhone-Connected Sleep Monitoring System

Apple recently acquired Beddit Sleep Monitor, an app and sleep system designed to monitor daily sleep habits through the iPhone, according to an updated privacy policy posted both on the Beddit website and in the Beddit app when creating an account. A link within the app also directs to the Apple Privacy Policy.

As of May 8, the Beddit privacy policy says the following:

Beddit has been acquired by Apple. Your personal data will be collected, used and disclosed in accordance with the Apple Privacy Policy.

The Beddit 3 Sleep Monitor, which can be purchased from Apple for $150, is a thin, flexible sensor that’s designed to be placed under the sheet on the top of a mattress. It collects and analyzes sleep-related data like sleep time and efficiency, heart rate, respiration, movement, snoring, room temperature, and room humidity.

All of the data collected by the Beddit Sleep Monitor is then made available to iPhone users through an accompanying Beddit iPhone app, which provides “personalized insights” and “customizable sleep coaching” to help users improve their sleep habits.

According to the Beddit website, the device uses ballistocardiography (BCG) to measure the mechanical activity of the heart, lungs, and other body functions. When the heart beats, for example, the Beddit sensor can measure the mechanical impulse generated by the acceleration of the blood through the circulatory system.

Based on the changes to the Beddit privacy policy, it appears Apple may have plans to continue selling the device and collecting health-related data from it, which could potentially be used to enhance future versions of products like the Apple Watch.

Tags: Apple acquisition, Beddit
Discuss this article in our forums

Google hopes new languages keep you from saying ‘adios’ to Allo

When Google released its Allo chat app late last year, the accompanying chatbot known as Google Assistant was only able to offer helpful advice and Smart Replies in English. Knowing full well that you can’t have an international chat assistant without supporting multiple languages, Google eventually introduced support for German, Hindi and Japanese as well. Today, that repertoire grows even larger as Assistant in Allo now supports two of the most popular languages in the world: French and Spanish.

Seeing as Google Assistant is the main way that Allo is differentiating itself from the other messaging apps out there, the support of multiple languages is pretty important if only to get additional marketshare. Allo users can use the Google Assistant to search for information, come up with quick smart replies, offer emoji suggestions or look up what the weather will be like that day. In order for Google Assistant to work in a different language, simply say “Talk to me in [language]” or “Speak to me in [language].”

It was just a couple of weeks ago that Google finally added basic chat functions to Allo, like encrypted incognito mode in group chats, link previews and the ability to backup and restore your messages. Still, we’re not sure if all of this will be enough to persuade folks to start using Allo over Hangouts. But at least now you can see if you can trick Google Assistant into offering “Me cago en la leche” as a Smart Reply.

Source: 9to5 Google