Nerdy extroverted software developer
4183 stories

DeepMind’s Go-playing AI doesn’t need human help to beat us anymore

1 Share

Google’s AI subsidiary DeepMind has unveiled the latest version of its Go-playing software, AlphaGo Zero. The new program is a significantly better player than the version that beat the game’s world champion earlier this year, but, more importantly, it’s also entirely self-taught. DeepMind says this means the company is one step closer to creating general purpose algorithms that can intelligently tackle some of the hardest problems in science, from designing new drugs to more accurately modeling the effects of climate change.

The original AlphaGo demonstrated superhuman Go-playing ability, but needed the expertise of human players to get there. Namely, it used a dataset of more than 100,000 Go games as a starting point for its own knowledge. AlphaGo Zero, by comparison, has only been programmed with the basic rules of Go. Everything else it learned from scratch. As described in a paper published in Nature today, Zero developed its Go skills by competing against itself. It started with random moves on the board, but every time it won, Zero updated its own system, and played itself again. And again. Millions of times over.

After three days of self-play, Zero was strong enough to defeat the version of itself that beat 18-time world champion Lee Se-dol, winning handily — 100 games to nil. After 40 days, it had a 90 percent win rate against the most advanced version of the original AlphaGo software. DeepMind says this makes it arguably the strongest Go player in history.

“By not using human data — by not using human expertise in any fashion — we’ve actually removed the constraints of human knowledge,” said AlphaGo Zero’s lead programmer, David Silver, at a press conference. “It’s therefore able to create knowledge itself from first principles; from a blank slate [...] This enables it to be much more powerful than previous versions.”

Silver explained that as Zero played itself, it rediscovered Go strategies developed by humans over millennia. “It started off playing very naively like a human beginner, [but] over time it played games which were hard to differentiate from human professionals,” he said. The program hit upon a number of well-known patterns and variations during self-play, before developing never-before-seen stratagems. “It found these human moves, it tried them, then ultimately it found something it prefers,” he said. As with earlier versions of AlphaGo, DeepMind hopes Zero will act as an inspiration to professional human players, suggesting new moves and stratagems for them to incorporate into their game.

As well as being a better player, Zero has other important advantages compared to earlier versions. First, it needs much less computing power, running on just four TPUs (specialized AI processors built by Google), while earlier versions used 48. This, says Silver, allows for a more flexible system that can be improved with less hassle, “which, at the end of the day, is what really matters if we want to make progress.” And second, because Zero is self-taught, it shows that we can develop cutting-edge algorithms without depending on stacks of data.

For experts in the field, these developments are a big part of what makes this new research exciting. That’s is because they offer a rebuttal to a persistent criticism of contemporary AI: that much of its recent gains come mostly from cheap computing power and massive datasets. Skeptics in the field like pioneer Geoffrey Hinton suggest that machine learning is a bit of a one-trick pony. Piling on data and compute is helping deliver new functions, but the current pace of advances is unsustainable. DeepMind’s latest research offers something of a rebuttal by demonstrating that there are major improvements to be made simply by focusing on algorithms.

“This work shows that a combination of existing techniques can go somewhat further than most people in the field have thought, even though the techniques themselves are not fundamentally new,” Ilya Sutskever, a research director at the Elon Musk-backed OpenAI institute, told The Verge. “But ultimately, what matters is that researchers keep advancing the field, and it's less important if this goal is achieved by developing radically new techniques, or by applying existing techniques in clever and unexpected ways.”

Professional 'Go' Player Lee Se-dol Set To Play Google's AlphaGo Photo: Google / Getty Images
An earlier version of AlphaGo made headlines when it beat Go champion Lee Se-dol in 2016. That version learned from humans how to play.

In the case of AlphaGo Zero, what is particularly clever is the removal of any need for human expertise in the system. Satinder Singh, a computer science professor who wrote an accompanying article on DeepMind’s research in Nature, praises the company’s work as “elegant,” and singles out these aspects.

Singh tells The Verge that it’s a significant win for the field of reinforcement learning — a branch of AI in which programs learn by obtaining rewards for reaching certain goals, but are offered no guidance on how to get there. This is a less mature field of work than supervised learning (where programs are fed labeled data and learn from that), but it has potentially greater rewards. After all, the more a machine can teach itself without human guidance, the better, says Singh.

“Over the past five, six years, reinforcement learning has emerged from academia to have much more broader impact in the wider world, and DeepMind can take some of the credit for that,” says Singh. “The fact that they were able to build a better Go player here with an order of magnitude less data, computation, and time, using just straight reinforcement learning — it’s a pretty big achievement. And because reinforcement learning is such a big slice of AI, it’s a big step forward in general.”

What are the applications for these sorts of algorithms? According to DeepMind co-founder Demis Hassabis, they can provide society with something akin to a thinking engine for scientific research. “A lot of the AlphaGo team are now moving onto other projects to try and apply this technology to other domains,” said Hassabis at a press conference.

Hassabis explains that you can think of AlphaGo as essentially a very good machine for searching through complicated data. In the case of Zero, that data is comprised of possible moves in a game of Go. But because Zero was not programmed to understand Go specifically, it could be reprogrammed to discover information in other fields: drug discovery, protein folding, quantum chemistry, particle physics, and material design.

Hassabis suggests that a descendant of AlphaGo Zero could be used to search for a room temperature superconductor — a hypothetical substance that allows electrical current to flow with zero lost energy, allowing for incredibly efficient power systems. (Superconductors exist, but they only currently work at extremely cold temperatures.) As it did with Go, the algorithm would start by combining different inputs (in this case, the atomic composition of various materials and their associated qualities) until it discovered something humans had missed.

 Photo by Sam Byford / The Verge
DeepMind co-founder Demis Hassabis says his company’s aim is to “solve intelligence.”

“Maybe there is a room temperature superconductor out and about. I used to dream about that when I was a kid, looking through my physics books,” says Hassabais. “But there’s just so many combinations of materials, it’s hard to know whether [such a thing exists].”

Of course, this would be much more complicated than simply pointing AlphaGo Zero at the Wikipedia page for chemistry and physics and saying “have at it.” Despite its complexity, Go, like all board games, is relatively easy for computers to understand. The rules are finite, there’s no element of luck, no hidden information, and — most importantly — researchers have access to a perfect simulation of the game. This means an AI can run millions of tests and be sure it’s not missing anything. Finding other fields that meet these criteria limits the applicability of Zero’s intelligence. DeepMind hasn’t created a magical thinking machine.

These caveats aside, the research published today does get DeepMind just a little bit closer to solving the first half of its tongue-in-cheek, two-part mission statement. Part one: solve intelligence; part two: use it to make the world a better place. “We’re trying to build general purpose algorithms and this is just one step towards that, but it’s an exciting step,” says Hassabis.

Read the whole story
Share this story

How to play YouTube in the background on Android and iOS

1 Comment

YouTube is the internet’s replacement for the TV, so it’s only natural that we’d use it as our background noise provider, however Google doesn’t make that easy on mobile devices. Unless you have a YouTube Red subscription, YouTube only plays as a foreground app, meaning you can’t use your phone and listen to music or podcasts on YouTube at the same time. Well, unless you cheat a little.

I stumbled upon a rather amusing new way to get YouTube videos playing in the background while testing the Google Pixel 2. The Android version of the Telegram messaging app can play YouTube video links directly, without needing to send you out to another app. This is all fine and conventional, but the fun starts when you minimize Telegram or even turn the screen off entirely. The audio from the YouTube content keeps going. With this little trick, I can just message myself anything I want to listen to (rather than watch) and leave it playing while I do other things with my phone. That’s especially kickass with one of the many 24-hour streaming “radio stations” on YouTube.

One discovery leads to another, and since I wasn’t too happy losing the use of my favorite messaging app, I had to look to see if there are other workarounds to Google’s prohibition on background YouTube play. Some phones like Samsung’s Galaxy Note 8 and LG’s V30 offer split-screen multitasking that will let you have a YouTube window alongside some other activity, but that’s obviously nowhere near as elegant as actually working in the background. As it turns out, people have already been quite industrious in figuring out alternative tactics.

On the iPhone (where the same Telegram trick doesn’t work), you can use a third-party browser such as Dolphin to start a YouTube video, then exit the app and resume playback via the iOS Control Center. Dead simple. Almost the same thing exists for Android users with Google’s own Chrome browser. If you switch to the desktop (rather than mobile) YouTube site in Chrome, you’ll be able to start a video in the browser, then exit the app and resume playback from the notification shade. This is by far the slickest of all options: you can play or pause either from notifications or from the phone’s lock screen. Even better: there’s nothing stopping you from opening further tabs in Chrome and continuing to use the browser as usual.

Here’s a handy list of steps to do it yourself, starting with Android:

  1. Open in the Chrome browser.
  2. In the three-dot menu at the top right, check the “Request desktop site” box.
  3. Navigate to the video you want to listen to and press play.
  4. If you’re shown a warning saying YouTube wants to send you notifications, accept it.
  5. Exit the Chrome app and you’ll be able to resume playback from the notification menu.

And on iOS:

  1. Download the Dolphin web browser to your iOS device.
  2. Open in Dolphin.
  3. Navigate to the video you want to listen to and press play.
  4. Exit the Dolphin app and you’ll be able to resume playback from the iOS Control Center.

It’s obvious Google knows that being able to just listen to audio from YouTube is a desired feature, which is why the company reserves it as a perk for YouTube Red subscribers. Browsing around online, you’ll also see quite a few other methods for enabling background play, some of which Google has already neutralized. But even so, the present workarounds for both Android and iOS users are laughably easy. Google would probably prefer that you don’t circumvent its systems, but hey, if it’s that easy to do, why not?

Read the whole story
Share this story
1 public comment
3 days ago
Here's another way: Subscribe to YouTube Red and just press home and it keeps playing like a normal music app.
Plano, Texas

Autonomous cars without human drivers will be allowed on California roads starting next year

1 Share

The California Department of Motor Vehicles will allow autonomous cars without steering wheels, foot pedals, mirrors, and human drivers behind the wheel to be tested on its roads starting next year. The department released its revised regulations today governing the safe deployment of autonomous vehicles on public roads. After a 15-day public comment period, those rules will be submitted to the state government, which will then begin enforcing them sometime in the middle of 2018.

California is an obvious hotbed for autonomous vehicle testing, so changes made to the state’s rules governing these tests are followed closely by companies like Google and Uber that are developing fleets of self-driving cars for public use. There are currently 42 companies testing some 285 autonomous vehicles that are licensed with the DMV, officials said. Nearly 1,000 safety drivers are licensed to test those vehicles, but after the state’s rules go into effect, companies would be allowed to deploy cars without any human behind the wheel.

The rules are a stark departure from the original proposal issued in 2015. Those draft regulations would have put the kibosh on autonomous vehicles without a human driver, such as Google’s steering wheel-less prototype. But then in May, the agency reversed itself, putting out a new proposal that would not only allow for the testing of self-driving cars without a human driver, but also regulate the manufacture and sale of fully autonomous vehicles.

Today, the DMV released a number of small revisions, including new provisions requiring manufacturers to notify local governments when they plan on testing autonomous vehicles in their cities or towns. The DMV is also issuing a new template for manufacturers to report the number of times the vehicle forced the human driver to take control because it couldn’t safely navigate the conditions on the road. And operators will be required to report to the state when some technological upgrades are made, such as a change in the level of automation.

The state is not changing its prohibition on the testing of autonomous trucks, arguing that a separate rule-making process will be needed to allow for the testing of self-driving vehicles over 10,000 pounds.

Congress is currently considering legislation that would allow companies to manufacture and deploy cars without traditional controls like pedals and steering wheels. The proposed bills would also preempt states from establishing their own laws overseeing autonomous testing, which could clash with California’s well-established system.

To be sure, this isn’t a clear path for fully autonomous vehicles to overrun the streets of California. Manufacturers would still need to receive approval or a waiver for exemption from the federal government before operating a vehicle on public roads without a human driver or conventional controls like a steering wheel or pedals.

But Congress is considering increasing federal motor vehicle safety standard (FMVSS) exemption caps from 2,500 to 100,000 — which is a wonky way of saying that it would allow automakers and tech firms to test (and eventually deploy) tens of thousands autonomous vehicles without components designed with humans in mind and required by federal safety standards. Right now, these companies are testing cars that can at best be considered Level 3 autonomous, meaning they still require some human intervention.

Read the whole story
Share this story

Microsoft finally admits Windows Phone is dead

1 Share

In a series of tweets, Microsoft’s Joe Belfiore has revealed that the software giant is no longer developing new features or hardware for Windows 10 Mobile. While Windows Phone fans had hoped Microsoft would update the platform with new features, it’s now clear the operating system has been placed into servicing mode, with just bug fixes and security updates for existing users.

“Of course we'll continue to support the platform.. bug fixes, security updates,” says Belfiore. “But building new features/hw aren't the focus.” During the recent Windows 10 Fall Creators Update development phase, it has been obvious Microsoft is no longer working on the mobile experience. The software maker branched its development process into “feature2,” leaving Windows Phone fans disappointed.

Microsoft’s Windows Phone platform has been dead for more than a year, but the company has never officially admitted it before. Microsoft gutted its phone business last year, resulting in thousands of job cuts. During Microsoft’s recent Build and Inspire conferences, CEO Satya Nadella dropped the company’s mantra of “mobile-first, cloud-first” in favor of a focus on what he describes as the intelligent cloud and intelligent edge. This new area of focus means Microsoft is now working on multi-device scenarios and cloud-powered technologies that don’t always involve Windows.

Belfiore also admits he switched to Android, just like Bill Gates, and that Microsoft will support Windows 10 customers who want to use Android and iOS on their phones. One of the big reasons Microsoft is giving up on Windows Phone is because developers never backed the platform. Microsoft had many problems with its Windows Phone app store, and the inability to retain apps once they were developed. “We have tried very hard to incent app devs,” explains Belfiore. “Paid money.. wrote apps 4 them.. but volume of users is too low for most companies to invest.”

It’s now clear Microsoft has faced the reality that people don't need Windows on their phones. The software maker has been focused on iOS and Android apps and experiences over the past couple of years. Microsoft’s Edge browser is making its way to iOS and Android, and the company seems to be focused on improving the experience of linking a phone to a PC and resuming apps and documents between the two.

A new Microsoft Launcher for Android even goes a step further in improving the overall Android experience, with clever customizations and PC sharing features. Expect to see a lot more of this in the future. It’s now unlikely we’ll ever see a Surface-branded phone or more Windows 10 Mobile-powered hardware. Microsoft’s investments in iOS and Android will only benefit Windows 10 customers, and it could make up for the company’s messy efforts in mobile over the years.

Read the whole story
Share this story

As US launches DDoS attacks, N. Korea gets more bandwidth—from Russia

1 Share

Enlarge / The North Korean border post for the Friendship Bridge, the rail link connecting North Korea to Russia. It now carries an Internet connection for North Korea to the rest of the world as well. (credit: DPRK KCNA)

As the US reportedly conducts a denial-of-service attack against North Korea's access to the Internet, the regime of Kim Jong Un has gained another connection to help a select few North Koreans stay connected to the wider world—thanks to a Russian telecommunications provider. Despite UN sanctions and US unilateral moves to punish companies that do business with the Democratic People's Republic of Korea, 38 North's Martyn Williams reports that Russian telecommunications provider TransTelekom (ТрансТелеКо́m) began routing North Korean Internet traffic at 5:30pm Pyongyang time on Sunday.

The connection, Williams reported, offers a second route for traffic from North Korea's Byol ("Star") Internet service provider, which also runs North Korea's cellular phone network. Byol offers foreigners in North Korea 1Mbps Internet access for €600 (US$660) a month (with no data caps).

Up until now, all Byol's traffic passed through a single link provided by China Unicom. But the new connection uses a telecommunications cable link that passes over the Friendship Bridge railway bridge—the only connection between North Korea and Russia. According to Dyn Research data, the new connection is now providing more than half of the route requests to North Korea's networks. TransTelekom (sometimes spelled TransTeleComm) is owned by Russia's railroad operator, Russian Railways.

Read 2 remaining paragraphs | Comments

Read the whole story
Share this story

Equifax CIO, CSO “retire” in wake of huge security breach

1 Share

Enlarge / A monitor displays Equifax Inc. signage on the floor of the New York Stock Exchange (NYSE) in New York, US, on Friday, Sept. 15, 2017. (credit: Michael Nagle/Bloomberg via Getty Images)

On Friday, Equifax announced that two top executives would be retiring in the aftermath of the company’s massive security breach that affected 143 million Americans.

According to a press release, the company said that its Chief Information Officer, David Webb, and Chief Security Officer, Susan Mauldin, would be leaving the company immediately and were being replaced by internal staff. Mark Rohrwasser, who has lead Equifax’s international IT operations, is the company’s new interim CIO. Russ Ayres, who had been a vice president for IT at Equifax, has been named as the company’s new interim CSO.

The notorious breach was accomplished by exploiting a Web application vulnerability that had been patched in early March 2017.

Read 3 remaining paragraphs | Comments

Read the whole story
Share this story
Next Page of Stories