Apple Launches New Blog to Share Details on Machine Learning Research
Apple today debuted a new blog called the “Apple Machine Learning Journal,” with a welcome message for readers and an in-depth look at the blog’s first topic: “Improving the Realism of Synthetic Images.” Apple describes the Machine Learning Journal as a place where users can read posts written by the company’s engineers, related to all of the work and progress they’ve made for technologies in Apple’s products.
In the welcome message, Apple encourages those interested in machine learning to contact the company at an email address for its new blog, machine-learning@apple.com.
Welcome to the Apple Machine Learning Journal. Here, you can read posts written by Apple engineers about their work using machine learning technologies to help build innovative products for millions of people around the world. If you’re a machine learning researcher or student, an engineer or developer, we’d love to hear your questions and feedback. Write us at machine-learning@apple.com
In the first post — described as Vol. 1, Issue 1 — Apple’s engineers delve into machine learning related to neural nets that can create a program to intelligently refine synthetic images in order to make them more realistic. Using synthetic images reduces cost, Apple’s engineers pointed out, but “may not be realistic enough” and could result in “poor generalization” on real test images. Because of this, Apple set out to find a way to enhance synthetic images using machine learning.
Most successful examples of neural nets today are trained with supervision. However, to achieve high accuracy, the training sets need to be large, diverse, and accurately annotated, which is costly. An alternative to labelling huge amounts of data is to use synthetic images from a simulator. This is cheap as there is no labeling cost, but the synthetic images may not be realistic enough, resulting in poor generalization on real test images. To help close this performance gap, we’ve developed a method for refining synthetic images to make them look more realistic. We show that training models on these refined images leads to significant improvements in accuracy on various machine learning tasks.
In December 2016, Apple’s artificial intelligence team released its first research paper, which had the same focus on advanced image recognition as the first volume of the Apple Machine Learning Journal does today.
The new blog represents Apple’s latest step in its progress surrounding AI and machine learning. During an AI conference in Barcelona last year, the company’s head of machine learning Russ Salakhutdinov provided a peek behind the scenes of some of Apple’s initiatives in these fields, including health and vital signs, volumetric detection of LiDAR, prediction with structured outputs, image processing and colorization, intelligent assistant and language modeling, and activity recognition, all of which could be potential subjects for research papers and blog posts in the future.
Check out the full first post in the Apple Machine Learning Journal right here.
Tag: machine learning
Discuss this article in our forums
Apple could let you discreetly call 911 with your fingerprint
Calling the emergency services is a deliberately easy process — just three digits in most countries, simple enough for a child to learn. But Apple may have plans to make it even more straightforward, after patenting a new way of calling for help using just your fingertips. The patent describes how the user could program different fingerprints, patterns and even the amount of force required to switch the phone into ‘panic mode’. This would result in the phone calling the emergency services, and could activate a live-stream of video, or provide location data via GPS. Personal safety smart jewelry — already on the market — works in a similar way.
Such technology could prove lifesaving in some situations, as demonstrated by a number of recent news reports. In London, a four-year-old boy saved his mother’s life with Siri’s help, and earlier this month New Mexico police reported that a smart home device had intervened in a domestic dispute. Whether Apple plans on doing anything with the patent, however, remains to be seen.
Via: Alphr
Source: United States Patent and Trademark Office
LastPass Families can manage passwords for a household of six
It’s hard enough keeping track of your own passwords, let alone your family’s. Unless you plan on outfitting your brood with biometric rings, your best bet is to sign up to a password managing tool. That way you can keep track of the logins your clan is racking up, and even reference them to create unique passwords in the future. Most popular services, such as 1Password, already offer family plans. And soon that small list will include LastPass too.
The password authenticator is launching LastPass Families later this summer. The plan will cover up to six members of a household. Once signed up, you’ll be able to organize info into an unlimited amount of folders. You can then open them up to other members. This allows you to keep certain data between you and your partner (like payment info for utility bills) while sharing passwords for streaming services with the kids. And, seeing as LastPass is free across devices, you’ll be able to manage your settings from anywhere.
There’s no mention of pricing yet. Although, LastPass will have to beat 1Password’s $5 a month family subscription to remain competitive. If you’re already forking out $12 annually for LastPass Premium, you’ll be given six months free access to LastPass Families when it lands. In the meantime, anyone can sign up for early access to the plan, considering you have a LastPass account.
Source: LastPass
Apple’s ‘Planet of the Apps’: lousy TV, good for developers
Ashley D’Arcy always dreamed of being on a reality TV show. But her job as a creative director for a chatbot doesn’t frequently put her in front of a camera. D’Arcy’s dream was realized, however, when she and dozens of other app makers were thrown into the spotlight on Apple’s first original TV series, Planet of the Apps (POTA). It’s been widely (and accurately) described as a cross between Shark Tank and The Voice, and it’s a tepid take on the high-stakes world of… app funding.
You may have heard how boring the show is, especially to my fellow tech reporters, who receive dozens of similar pitches every day. POTA is disappointing to those who expect more from the company that’s basically responsible for today’s thriving app ecosystem. You would think that Apple would be able to create a show that’s unique, informative and entertaining, given its expertise on the subject matter and history of innovating (even if its latest projects have been bland).

But the most interesting part of the show barely gets enough screen time. It’s a clever spin on the elevator pitch — a physical escalator that takes hopeful participants on a 60-second ride from the developer lounge while they describe their app. At the bottom, they face the celebrity judges: Jessica Alba, Gwyneth Paltrow, Garry Vaynberg and will.i.am. (I guess Jony Ive was too busy designing the next iPhone.) The judges swipe yes or no (green or red) on their iPads after the escalator pitch, and if at least one of them says yes, the developers get to explain their product in more detail. At the end of that, all four judges get to decide if they want to help out the app, and the participants pick the mentor they prefer. Sound familiar? That’s because it’s the basic format for The Voice.
After the escalator pitch, the rest of the show follows typical reality TV tropes. Over the course of about 50 minutes, you’ll watch the developers work with their mentors to refocus their apps and come up with a killer pitch that will convince Lightspeed Venture Partners to give them money. Basically, after snagging a mentor on The Voice for apps, they’ll be working to get funding from Shark Tank for apps. Yup. Way to repurpose existing formats, Apple.
For a show about developers, there is shockingly little coding or app design on-screen. What you do get is a pop-up label every time a barely jargony phrase is dropped, explaining what the term means. It’s strangely condescending and seems somewhat misguided. The show targets a mainstream audience that might not get terms like API, SDK or UX, and for these folks those labels make sense. The thing is, because of its subject matter, Planet is theoretically more likely to appeal to those in the tech industry who already know the majority of those words.

But POTA is not completely without merit. At the least, it helps the participating companies. On last week’s episode, I watched as three apps that I would normally have written off — Poncho, Cheddar and Stop, Breathe & Think (SBT) — refocused and gained serious potential and funding.
Participants agreed that pitching a venture capital fund in real life isn’t much different from the way it plays out in POTA, barring the cameras and production schedules. “The only difference is, the VCs made their decision within probably 10 to 15 minutes, versus the real world where they probably take weeks, months before they can commit,” Poncho’s founder Kuan Huang told Engadget. “I’m pretty sure Lightspeed did lots of homework before they even talked to us.” Otherwise, Huang noted that the back and forth over details like metrics and financials very much resembled a venture capital pitch meeting outside the show, except Lightspeed was better prepared.
Although there wasn’t a significant difference between being on the show and fund-raising in real life, the companies involved still benefited to varying degrees. Of the trio from last week’s episode, Poncho gained the most. It’s a messaging bot in the form of a Brooklyn cat that tells you the weather in a personable way. I was already aware of Poncho before the episode, and while I found the character’s friendly persona interesting, I wasn’t as convinced by the promise of yet another chatbot.
After Poncho was picked by Paltrow, Huang and D’Arcy talked to Giphy founder Alex Chung during a six-week incubation period and refocused their objective with Chung’s perspective and input. They went to Lightspeed with a compelling pitch — Poncho would be a content platform that served up intelligent alerts, schedules, news and even buying suggestions to your phone’s lock screen. Because of the work by D’Arcy and her team, Poncho is humorous, charming and adorable, which Lightspeed believed would resonate with the vast market of millennials today.
As they went over the details of the pitch, Lightspeed also pointed out that Poncho’s user base had a significant portion of younger women, whom they said were tastemakers. In fact, they said, Snapchat had similar demographics before it blew up. Lightspeed would know too: It was one of the first investors in the ephemeral messaging app.
In just one episode (that took six months to produce), Poncho went from a simple chat service looking to monetize its product to a content platform that aspires to be as big as Snapchat. I sure am paying attention now.

Stop, Breathe & Think saw more-quantifiable improvements. CEO Julie Campistron told Engadget that since being featured in the app store after the show, downloads of the company’s app have doubled. Being on POTA also sped up the process of creating a child-friendly version of Stop, Breathe and Think’s meditation program, Campistron said. That made-for-kids app has seen a 60 percent increase in downloads since the episode aired. Thanks to the show, she was also able to conduct focus groups (set up by the production company) to learn how children respond to her app.
For Cheddar’s CEO Jon Steinberg, having to repeatedly describe his business in a short time made him better at it. “The challenge of pitching on the escalator made me have to refine the idea in a very specific and simple way,” he said. “It forced me to have to explain the company in 60 seconds.” Steinberg ended up securing $2.5 million from Lightspeed to revamp his app (it’s worth noting Lightspeed had already invested in Cheddar prior to the show). Plus, he got to spend time with his mentor will.i.am, who Steinberg said was his first choice even before filming started, and they remain in touch.
Ultimately, POTA is a barely entertaining program that really only benefits its participants. The thing is, Apple had almost nothing to do with the improvements each company in the most recent episode saw, aside from download gains from being featured in the app store. It would have been nice to see the iPhone maker’s own developers weigh in on what makes an app successful or go full geek on the benefits of one programming language over another. If presented differently, maybe POTA would have a chance with app nerds and the TechCrunch crowd. But as it is, the show lacks punch and flavor — essential ingredients for drawing in a mainstream audience. If this is an indicator of what to expect from Apple’s upcoming lineup of original content, we may be in for more disappointment yet.
Four Major iPhone Suppliers Join Apple in Countersuit Against Qualcomm
Compal, Hon Hai Precision/Foxconn, Pegatron, and Wistron have filed a countersuit against iPhone LTE modem supplier Qualcomm in an attempt to prevent Qualcomm from successfully forcing them to pay certain licensing fees related to the iPhone’s assembly (via Bloomberg). The move is a response to a lawsuit from May when Qualcomm sued the four suppliers for “breaching their license agreements” by failing to pay royalties on the use of Qualcomm’s technology in the assembly of Apple’s devices.
Now, in a court filing today, the four companies have claimed that Qualcomm is asking for payments “massively in excess” of what it would normally receive. If the countersuit is successful, Apple said that it could cost Qualcomm billions in refunded fees and damages. For the manufacturers’ part, the companies described the Qualcomm suit as “yet another…anticompetitive scheme” by Qualcomm.
According to The Wall Street Journal, Apple is said to be covering the legal fees associated with the four manufacturers’ defense, and that it would soon file a separate motion to combine the new countersuit with its own suit against Qualcomm, creating one unified case.
Apple’s key contention is that Qualcomm is asking the court to force the contract manufacturers to pay licensing fees due on iPhones above the level the chipmaker normally receives.
The manufacturers — Compal, Hon Hai Precision and its Foxconn subsidiary, Pegatron Corp., and Wistron Corp. — denied violating any payment agreements. They called the Qualcomm suit against them “yet another chapter of Qualcomm’s anticompetitive scheme to dominate modem chip markets, extract supracompetitive royalties, and break its commitments to license its cellular technology on fair, reasonable and non-discriminatory terms.”
Apple and its manufacturing partners have also responded to a separate Qualcomm court filing, in which the LTE modem supplier requested an injunction to force Apple’s iPhone manufacturers to keep paying royalties during the legal battle. Apple, Compal, Foxconn, Pegatron, and Wistron have objected to the request, stating that “there’s no harm to Qualcomm waiting to get paid” until the end of the case, when the court determines the correct amount.
Earlier this week, Qualcomm CEO Steve Mollenkopf said that an out of court settlement for the case could still happen, although he had no specific confirmation or update on the legal battle that suggested a settlement was coming between Qualcomm and Apple. The two companies have been embroiled in the court case since January, sparked by an FTC complaint about Qualcomm’s anticompetitive patent licensing practices, and continued with Apple’s own lawsuit against the supplier, and then Qualcomm’s countersuit response.
If there is no settlement between Apple and Qualcomm, the case is expected to continue for the next few years.
Tag: Qualcomm
Discuss this article in our forums
Apple Resumes Selling Nokia’s Digital Health Products Following Settled Legal Dispute
Apple today resumed carrying Nokia’s digital health products on its online store in the United States and Canada, nearly two months after reaching a business cooperation agreement with the Finnish company.
The products available to purchase again include Nokia’s Body Cardio Scale, Body+ Scale, BPM+ Blood Pressure Monitor, and Thermo Thermometer. Each of the accessories were previously sold under Withings, but Nokia retired the Withings brand last month, roughly one year after acquiring the French company.
Apple’s website indicates the products are also available for pickup at Apple’s retail stores between Friday, July 21 and Monday, July 24.
Apple had removed all Withings accessories from its online and retail stores around the world in December due to a legal dispute. On May 23, however, Apple and Nokia announced they settled all litigation related to their intellectual property dispute and agreed to a multi-year patent license.
The legal dispute began last December, when Nokia filed dozens of patent infringement lawsuits against Apple in the United States and other countries. Apple countersued Nokia, accusing the former phone giant of transferring patents to patent holding entities to squeeze additional royalties from the iPhone maker.
As part of the settlement, Nokia will receive an up-front cash payment from Apple, with additional revenues during the term of the agreement. Nokia will also be providing “certain network infrastructure product and services” to Apple.
Tags: Withings, Apple retail, Nokia
Discuss this article in our forums
Samsung Debuts English-Language Voice Capabilities for Siri Rival ‘Bixby’ Following Initial Delay
When the Galaxy S8 and Galaxy S8+ smartphones went on sale in April, voice support for Bixby in the United States was missing due to Samsung’s decision to delay the virtual assistant’s English-language launch until what was then estimated to be sometime in late spring. This week, the company has announced that voice capabilities for Bixby are now arriving for Galaxy S8 and S8+ users across the United States.
Samsung describes Bixby as “fundamentally different” than AI helpers like Siri and Microsoft’s Cortana because of its deeper and more nuanced integration into the Galaxy’s core apps, as well as some third party apps at launch including Google Maps, Google Play Music, YouTube, and Facebook. The Bixby integration into these third party apps is available through a service called Bixby Labs, which Galaxy S8 owners can opt into through their smartphone.
Bixby’s many features include letting users complete simple tasks like turning on the device’s flashlight and taking a screenshot or selfie. More complex flows can be arranged as well, like asking the assistant to gather all of the photos taken over the past week into one album called “Vacation,” and then sending it to family and friends. Samsung plans to continuously update Bixby with new features, app support, languages and devices, and the company said that thanks to deep learning, the assistant will only improve over time.
What’s more, because Bixby is deeply integrated into the device’s operating system – rather than being a separate app –users can seamlessly switch between controlling an app via voice or via touch commands, rather than choose one or the other ahead of time.
“There are over 10,000 functions on our smartphones, but each day, people may use less than five percent of them. The features on your phone are only useful if you know how to find them. Our goal with Bixby is to make it easier to use our phones, creating the best possible user experience for our customers. That’s why we created Bixby – an intuitive new way to do more things with your phone,” says Injong Rhee, Executive Vice President, Head of R&D, Software and Services of Mobile Communications Business, Samsung Electronics.
With the Bixby update, Galaxy S8 users can simply press the dedicated Bixby button on their device to call up the assistant and begin asking it questions and performing tasks, and a voice activated “Hi, Bixby” feature is also available. Right now, Bixby recognizes English and Korean languages, but Samsung noted that, “Not all accents, dialects and expressions” will be recognized.
Every time that Galaxy S8 users call up Bixby they will also earn experience points in a gamification system that Samsung is also rolling out for the launch, and ending September 14, 2017. Each message sent, call made, setting updated, or any action performed through Bixby will earn XP, which users can convert into Samsung Rewards points that can be amassed to try to win bigger prizes like Samsung products, gift cards, trips, and more.

Earlier this month, it was rumored that Samsung is planning to enter the smart speaker market with a device that uses Bixby as a way for users to interact with music playback. Codenamed “Vega,” the project would place Samsung as a competitor against Apple in yet another category, since the Cupertino company at WWDC this year announced the December launch of the HomePod, its Siri-powered speaker.
Delays related to Bixby’s U.S. launch are reportedly behind a slowdown on the smart speaker project, and a report out of The Korea Herald this week has furthermore placed the launch of a Samsung speaker well into the future. Sources familiar with the matter claimed that Samsung is “not enthusiastic” about a smart Bixby speaker because it “does not view Al speakers as marketable” at this time, thanks to the domination of products like Amazon Echo. Rather than dive into an uncertain market, Samsung is said to be taking a wait-and-see approach for the potential launch of its Bixby speaker.
Tags: Samsung, Bixby
Discuss this article in our forums
A security robot called Steve rolled into a pool and drowned
Why it matters to you
It looks like it’ll be a while before we can fully rely on security robots to keep us safe.
The robot apocalypse is clearly a ways off if news out of Washington, D.C. this week is anything to go by.
A security robot working at the Washington Harbour complex in Georgetown met a nasty end when it toppled into a water feature and promptly drowned.
Steve, as this particular Knightscope K5 robot is affectionately known, had only been patrolling his patch since last week but, for reasons currently unknown, ended up taking an unexpected dip in the water, rendering itself utterly useless.
Images posted on social media showed a rather sorry sight, with Steve face down in the fountain, floating in a very unrobot-like way as staff attempted to lift the lifeless bot from its watery grave.
It’s a fun day here at @gmmb. The super high-tech security robot at our office complex has had a mishap. pic.twitter.com/nhRshrJA9w
— Greg Pinelo (@gregpinelo) July 17, 2017
DigitalTrends’ Kyle Wiggers, who recently met one of Knightscope’s 6-foot, 400-pound K5 robots, described it as “something like a mix between a Dalek from Doctor Who and Eve from Wall-E.” Designed to work alongside human security personnel rather than replace them, the wheel-based K5 comes packed with sensors and cameras designed to help it make sense of its surroundings while at the same time alerting its human counterparts if it spots any suspicious behavior or dodgy characters.
The operator of Washington Harbour was evidently stoked to have Steve keeping an eye on the complex, posting a message on Facebook last week about the “new sheriff in town.”
It aded, “We’re super excited to announce our new test pilot security technology … This bot is making his debut at D.C. headquarters this week and has an extensive catalog of security capabilities. He’s currently mapping out the grounds here to be fully autonomous and ready to launch in the upcoming days.”
While Steve’s mapping technology obviously needs some tweaking, Knightscope chose to see the funny side of this week’s mishap.
BREAKING NEWS: “I heard humans can take a dip in the water in this heat, but robots cannot. I am sorry,” said K5 in an official statement. pic.twitter.com/nWC4tubv9w
— Knightscope (@iKnightscope) July 18, 2017
It’s not the first time the K5 has unintentionally landed up in the news. Just a few months ago a man apparently assaulted the robot in a Silicon Valley parking lot. Despite being knocked to the ground, the K5 did what it was supposed to do and alerted cops of the assault. The alleged assailant was later arrested.
As for Steve, there’s no word about whether he’s going to be fixed, or indeed if the folks at Washington Harbour will be asking for a replacement anytime soon, though if they do, we suggest they erect a barrier around the fountain beforehand. Either that or teach the K5 to recognize water.
Google’s mobile search app now has a personalized news feed
Google launches a news feed of its own within its mobile search app.
Google is increasingly looking to make its products more social, and its latest attempt is to create a feed within its mobile search app that will show you news stories, sports scores, and videos, all based on your previous searches. The company introduced news stories in the Google app back in December, and today’s update sees Google building upon that by leveraging its AI smarts.

The feed includes topics currently trending in your area, and Google also pulls in information from other services, including Google Maps, Gmail, and YouTube searches. For instance, if you’re interested in Game of Thrones, Google will serve up news articles based on HBO’s hit show. Or if you’re going to vacation to a particular country, Google will offer suggestions on places to visit.
The goal is to turn the default Google app into a destination for news and entertainment, much like Facebook. The feed is now rolling out in the U.S., and will be making its way to other countries over the coming weeks.
Deal: Save 20% off your first six months of cell service at TPO

If you’re looking for a sweet deal on a new phone plan, look no further than TPO.
TPO (The People’s Operator) — an alternative carrier that runs off Sprint and T-Mobile in the U.S. — is currently offering 20% off your first six months of service with new activations. You can activate a new number and a new device, or keep your current setup, to save on their 500MB, 1GB or 2.5GB plans. The best part is that 10% of your monthly spend goes to a charity of your choice!
Here’s what TPO has to offer:
- 500MB plan with 1000 min of talk and unlimited text, pay $12.80/month for the first 6 months and then $16/month thereafter.
- 1GB plan with unlimited talk & text, pay $16.80/month for the first 6 months and then $21/month thereafter.
- 2.5GB plan with unlimited talk & text to US, Mexico & Canada, pay $31.20/month for the first 6 months and then $39/month thereafter.
So if you’re looking for a new carrier, or need another line that won’t break the bank, this offer is quite a deal. TPO will let you save a few bucks off your first six months of service, helping to ease your way into a new plan. TPO supports all Sprint- and T-Mobile-compatible devices, and there are no contracts or data overages. You’ll also be able to easily give back to a bunch of great causes.
See at TPO



