All posts in “Artificial Intelligence”

Here’s everything Adobe announced at its MAX conference this week

Adobe is hosting its MAX conference this week in Las Vegas, and even though the company has long moved to a subscription model with regular updates, it still uses the event to launch its most important updates and new products. Here is our rundown of the biggest announcements, which range from new tools for designers, a completely new version of Lightroom and a few crazy experiments.

Apple responds to Senator Franken’s Face ID privacy concerns


Apple has now responded to a letter from Senator Franken last month in which he asked the company to provide more information about the incoming Face ID authentication technology which is baked into its top-of-the-range iPhone X, due to go on sale early next month.

As we’ve previously reported, Face ID raises a range of security and privacy concerns because it encourages smartphone consumers to use a facial biometric for authenticating their identity — and specifically a sophisticated full three dimensional model of their face.

And while the tech is limited to one flagship iPhone for now, with other new iPhones retaining the physical home button plus fingerprint Touch ID biometric combo that Apple launched in 2013, that’s likely to change in future.

After all, Touch ID arrived on a single flagship iPhone before migrating onto additional Apple hardware, including the iPad and Mac. So Face ID will surely also spread to other Apple devices in the coming years.

That means if you’re an iOS user it may be difficult to avoid the tech being baked into your devices. So the Senator is right to be asking questions on behalf of consumers. Even if most of what he’s asking has already been publicly addressed by Apple.

Last month Franken flagged what he dubbed “substantial questions” about how “Face ID will impact iPhone users’ privacy and security, and whether the technology will perform equally well on different groups of people”, asking Apple for “clarity to the millions of Americans who use your products” and how it had weighed privacy and security issues pertaining to the tech itself; and for additional steps taken to protect users.

Here’s the full list of 10 questions the Senator put to the company:

1.      Apple has stated that all faceprint data will be stored locally on an individual’s device as opposed to being sent to the cloud.

a.      Is it currently possible – either remotely or through physical access to the device – for either Apple or a third party to extract  and obtain usable faceprint data from the iPhone X?

b.      Is there any foreseeable reason why Apple would decide to begin storing such data remotely?

2.     Apple has stated that it used more than one billion images in developing the Face ID algorithm. Where did these one billion face images come from?

3.     What steps did Apple take to ensure its system was trained on a diverse set of faces, in terms of race, gender, and age? How is Apple protecting against racial, gender, or age bias in Face ID?

4.     In the unveiling of the iPhone X, Apple made numerous assurances about the accuracy and sophistication of Face ID. Please describe again all the steps that Apple has taken to ensure that Face ID can distinguish an individual’s face from a photograph or mask, for example.

5.     Apple has stated that is has no plans to allow any third party applications access to the Face ID system or its faceprint data. Can Apple assure its users that it will never share faceprint data, along with the tools or other information necessary to extract the data, with any commercial third party?

6.      Can Apple confirm that it currently has no plans to use faceprint data for any purpose other than the operation of Face ID?

7.     Should Apple eventually determine that there would be reason to either begin storing faceprint data remotely or use the data for a purpose other than the operation of Face ID, what steps will it take to ensure users are meaningfully informed and in control of their data?

8.      In order for Face ID to function and unlock the device, is the facial recognition system “always on,” meaning does Face ID perpetually search for a face to recognize? If so:

a.      Will Apple retain, even if only locally, the raw photos of faces that are used to unlock (or attempt to unlock) the device?

b.      Will Apple retain, even if only locally, the faceprints of individuals other than the owner of the device?

9.      What safeguards has Apple implemented to prevent the unlocking of the iPhone X when an individual other than the owner of the device holds it up to the owner’s face?

10.   How will Apple respond to law enforcement requests to access Apple’s faceprint data or the Face ID system itself?

In its response letter, Apple first points the Senator to existing public info — noting it has published a Face ID security white paper and a Knowledge Base article to “explain how we protect our customers’ privacy and keep their data secure”. It adds that this “detailed information” provides answers “all of the questions you raise”.

But also goes on to summarize how Face ID facial biometrics are stored, writing: “Face ID data, including mathematical representations of your face, is encrypted and only available to the Secure Enclave. This data never leaves the device. It is not sent to Apple, nor is it included in device backups. Face images captured during normal unlock operations aren’t saved, but are instead immediately discarded once the mathematical representation is calculated for comparison to the enrolled Face ID data.”

It further specifies in the letter that: “Face ID confirms attention by directing the direction of your gaze, then uses neural networks for matching and anti-spoofing so you can unlock your phone with a glance.”

And reiterates its prior claim that the chance of a random person being able to unlock your phone because their face fooled Face ID is approximately 1 in 1M (vs 1 in 50,000 for the Touch ID tech). After five unsuccessful match attempts a passcode will be required to unlock the device, it further notes.

“Third-party apps can use system provided APIs to ask the user to authenticate using Face ID or a passcode, and apps that support Touch ID automatically support Face ID without any changes. When using Face ID, the app is notified only as to whether the authentication was successful; it cannot access Face ID or the data associated with the enrolled face,” it continues.

On questions about the accessibility of Face ID technology, Apple writes: “The accessibility of the product to people of diverse races and ethnicities was very important to us. Face ID uses facial matching neural networks that we developed using over a billion images, including IR and depth images collected in studies conducted with the participants’ informed consent.”

The company had already made the “billion images” claim during its Face ID presentation last month, although it’s worth noting that it’s not saying — and has never said — it trained the neural networks on images of a billion different people.

Indeed, Apple goes on to tell the Senator that it relied on a “representative group of people” — though it does not confirm exactly how many individuals, writing only that: “We worked with participants from around the world to include a representative group of people accounting for gender, age, ethnicity and other factors. We augmented the studies as needed to provide a high degree of accuracy for a diverse range of users.”

There’s obviously an element of commercial sensitivity at this point, in terms of Apple cloaking its development methods from competitors. So you can understand why it’s not disclosing more exact figures. But of course Face ID’s robustness in the face of diversity remains to be proven (or disproven) when iPhone X devices are out in the wild.

Apple also specifies that it has trained a neural network to “spot and resist spoofing” to defend against attempts to unlock the device with photos or masks. Before concluding the letter with an offer to brief the Senator further if he has more questions.

Notably Apple hasn’t engaged with Senator Franken’s question about responding to law enforcement requests — although given enrolled Face ID data is stored locally on a user’s device in the Secure Element as a mathematical model, the technical architecture of Face ID has been structured to ensure Apple never takes possession of the data — and couldn’t therefore hand over something it does not hold.

The fact Apple’s letter does not literally spell that out is likely down to the issue of law enforcement and data access being rather politically charged.

In his response to the letter, Senator Franken appears satisfied with the initial engagement, though he also says he intends to take the company up on its offer to be briefed in more detail.

“I appreciate Apple’s willingness to engage with my office on these issues, and I’m glad to see the steps that the company has taken to address consumer privacy and security concerns. I plan to follow up with Apple to find out more about how it plans to protect the data of customers who decide to use the latest generation of iPhone’s facial recognition technology,” he writes.

“As the top Democrat on the Privacy Subcommittee, I strongly believe that all Americans have a fundamental right to privacy,” he adds. “All the time, we learn about and actually experience new technologies and innovations that, just a few years back, were difficult to even imagine. While these developments are often great for families, businesses, and our economy, they also raise important questions about how we protect what I believe are among the most pressing issues facing consumers: privacy and security.”

These AI learn by competing against each other, and it looks ridiculous

Researchers at Elon Musk’s startup, OpenAI, think they have discovered the most efficient way to train artificial neural networks: have them compete against each other.

The researchers have created simulations of human sports, like sumo wrestling and soccer, and pitted the AI players against each other. Winning techniques are rewarded through behavioral reinforcement, while the losing AI is encouraged to try different moves.

The ‘self-play’ method ensures that the AI are always learning, and that their tasks are always the perfect difficulty: facing off against different versions of itself. And, yes, it looks pretty funny.

Huawei Mate 10 Pro is an AI-powered phone with a unique design

China’s top smartphone maker, Huawei, has recently overtaken Apple in global smartphone sales and it’s second just behind Samsung. 

But with their new, most ambitious smartphone yet — the Mate 10 — the Chinese company is finally ready to take on the competition from all sides. 

The first thing that really strikes you about the Mate 10 and its more powerful variant, the Mate 10 Pro, is the snazzy, sleek, compact design, particularly compared with the past models.

Both the Mate 10 and Mate 10 Pro have an aluminium frame, glass back and front which is slightly bent around the edges and really slim bezels on top and bottom.

On the back, the Leica-branded vertical dual-camera module is encased in a horizontal strip that is in a lighter colour compared to the rest of the phone.

In the Pro version that strip, situated just above the fingerprint sensor and combined with the almost bezel-less screen, gives the phone a unique flavour, putting it on par with the new Note 8 or the LG V30. 

Image: huawei

Another important development on the Pro is the massive 6-inches, 18:9 display with 2160×1080 resolution and — finally — OLED technology. Until now, Huawei had obstinately refused to improve full HD, which is still the screen resolution of the Mate 10 version. 

But with the OLED screen, the Mate Pro is telling the iPhone and others things are getting serious. 

In September, after Apple launched its brand-new iPhone X with a much-talked about facial recognition feature, Huawei mocked the American tech giant with a Facebook video which branded their new device #TheRealAiPhone. 

And certainly, in the pre-briefing attended by Mashable, the Mate 10’s AI features took the biggest part of the Chinese company’s PR effort. So what’s all the fuzz about? 

The Mate 10 Pro will have Huawei’s new Kirin 970 chipset and according to Huawei it will be the “world’s 1st smartphone with dedicated neural network processor unit”. 

The chipset has a dedicated AI chip, called NPU or Neural Processing Unit, which is able to simultaneously process complex computing while interacting with its surroundings. Huawei claims the NPU has 25% improvement on performance and 50% on efficiency compared to the normal CPU chip. 

A chart produced by Huawei showed that in a minute the Mate Pro is able to recognise 2,005 photos, while the iPhone 7 Plus 487 and the Samsung S8 just 95. 

Among other things, this means the AI engine is able to give you smart tips based on the hour of the day, prioritising apps you’re more likely to use for example, or switching to low-light or eye-care mode. 

Image: HUAWEI

With Microsoft’s translation app, pre-loaded in the phone, the NPU is able to give fast real-time translation replacing the original text in augmented reality mode while pointing at anything that needs translation. 

But it’s with the camera — Huawei calls it “intelligent photography” — that the NPU has a vast area of applications. 

The phone replicates the 12-megapixel colour/20-megapixel monochrome combination of the Mate 9, with Bokeh effect, digital zoom and optical image stabilization. 

This time, though, those first-class Leica optics will have an impressive f/1.6 lens aperture, just like the LG V30, meaning that in low-light situations the camera will perform better than the f/1.8-equipped iPhone 7, allowing 25% more light. 

Huawei prides itself of the AI-powered object-recognition function that lets the camera identify objects in real-time so that it can swiftly change its settings and adjusts its metrics accordingly. 

At the moment, the categories are: snow, food, sunset, cat, dog, flower, plant and portrait, but Huawei claims the machine-learning system will add more categories as they become available. 

It is not clear whether these functions will overall improve your pictures or they’re just gimmicky.

Mashable briefly tried out the phone’s AI-fuelled camera system and while some features are actually useful — it seemed to swiftly and simultaneously detect night mode + portrait + words, and automatically adjust hue, temperature and sharpness — in other examples, such as the distinction between plant vs flower, it all seemed a bit preposterous. 

Image: HUAWEI

A professional photographer in the room explained how the camera’s intelligent core can help you shoot more accurate pictures of a green field at sunset, for example, as the colour “green” notoriously messes with your camera settings. Only time and a more throughout review will tell the truth. 

The Mate 10 will be available in four colours — Mocha Brown, Champagne Gold and Pink Gold — while the Mate 10 Pro is available in: Midnight Blue, Mocha Brown, Titanium Gray and Pink Gold. 

The Pro will have 6GB RAM + 128GB of memory while the Mate 10 will have 4GB RAM and 64GB memory. Both will have a 4,000mAh battery.

There will also be an exclusive Porsche design variant in Diamond Black. 

The Mate 10 costs 699 euros while the Pro will be at 799 euros. The Porsche design is at a staggering 1395 euros. 

Https%3a%2f%2fvdist.aws.mashable.com%2fcms%2f2017%2f10%2f601f02c0 6f8c 60be%2fthumb%2f00001

China’s biggest messaging app accidentally translates ‘black foreigner’ to the N-word

Image: HOW HWEE YOUNG/EPA/REX/Shutterstock

WeChat, China’s biggest messaging app, has apologised for a gaffe, where it translated the phrase “black foreigner” to the N-word.

The mistake was first spotted by Ann James, an American living in Shanghai. She translated an incoming Chinese message into English, which produced the text: “The n****r’s still late.”

The original Chinese message used a more neutral term, hei laowai, or “black foreigner.”

(Editor’s note: The language in the screenshot below has been obscured due to its offensive nature)   

WeChat rectified the error within 24 hours, but the company told Chinese outlet Sixth Tone that the translation was based on its neural machine learning engine, that picked up the term from broader usage.

WeChat, which has a huge base of over 900 million users, has been able to translate messages in-app since 2014. It relies on a combination of translation sources, including its own AI engine, and third parties like Microsoft Translator.

The error is reminiscent of other translation engines that have tried to learn from analysing big data. In August, two Chinese chatbots — one created by Microsoft — were taken down after they started posting unpatriotic content about the government.

Last year, another Microsoft chatbot, Tay, was pulled after it started tweeting racist and crude messages.

Https%3a%2f%2fblueprint api production.s3.amazonaws.com%2fuploads%2fvideo uploaders%2fdistribution thumb%2fimage%2f82043%2fcaef07cc a17a 4d83 8efb 23ebb74c68b7