This work presents a fast and non-convex algorithm for robust subspace recovery. The data sets considered include inliers drawn around a low-dimensional subspace of a higher dimensional ambient space, and a possibly large portion of outliers that do not lie nearby this subspace. The proposed algorithm, which we refer to as Fast Median Subspace (FMS), is designed to robustly determine the underlying subspace of such data sets, while having lower computational complexity than existing methods. We prove convergence of the FMS iterates to a stationary point. Further, under a special model of data, FMS converges to a point which is near to the global minimum with overwhelming probability. Under this model, we show that the iteration complexity is globally bounded and locally $r$-linear. The latter theorem holds for any fixed fraction of outliers (less than 1) and any fixed positive distance between the limit point and the global minimum. Numerical experiments on synthetic and real data demonstrate its competitive speed and accuracy.

Sales of drones are soaring - and so are the concerns about how easily these devices can be hacked. Researchers have discovered that manufacturers are unknowingly leaving digital doors unlocked that give hackers an opportunity to wreak havoc on the system. The team found three different vulnerabilities in the devices that allowed them to send rogue commands from a laptop to disrupts its normal operation, land it or send it crashing to the ground. Researchers have discovered that manufacturers are unknowingly leaving digital doors unlocked that give hackers an opportunity to wreak havoc on the system. The students performed wireless network penetration testing on a popular hobby drone and developed'exploits' from the vulnerabilities found to disrupt the process of operators to control flights.

Chances are, you're already hip-deep in machine-learning applications. It's how Google Photo organizes those pictures from your vacation in Spain. It's how Facebook suggests tags for the pictures you took at last week's soccer match. It's how the cars of nearly every major automaker can help you avoid unsafe lane changes. Machine learning – which enables a computer to learn without new programming – is exploding in its ability to handle highly complex tasks.

Cylance, an Irvine, California-based cybersecurity startup that taps machine-learning and artificial intelligence (A.I.) to thwart malware, has raised 100 million in a Series D round led by Blackstone Tactical Opportunities and Insight Venture Partners, with participation from existing investors. The company was founded in 2012 by Stuart McClure, who sold an Internet security firm to McAfee for 86 million eight years ago and came on board as McAfee's chief technology officer (CTO). Prior to this round, Cylance had already raised about 77 million, including a 42 million round last July and 20 million the previous year. McClure has an interesting background -- his venture into the security realm was influenced by a deadly plane crash he was involved in back in 1989, one which resulted in the death of nine passengers. The event was apparently caused by a flaw in the Boeing 747's locking mechanism -- a known fault that the company ignored.

Academics from the Future of Humanity Institute (FHI), part of the Oxford Martin School, are teaming up with Google DeepMind to make artificial intelligence safer. Stuart Armstrong, the Alexander Tamas Fellow in Artificial Intelligence and Machine Learning at FHI and Laurent Orseau, of Google DeepMind, will present their research on reinforcement learning agent interruptibility at the UAI 2016 conference in New York City later this month. Orseau and Armstrong's research explores a method to ensure that reinforcement learning agents can be repeatedly safely interrupted by human or automatic overseers. This ensures that the agents do not "learn" about these interruptions, and do not take steps to avoid or manipulate the interruptions. Interruptibility has several advantages as an approach over previous methods of control.

A lot can change in the space of 10 years: At the 2006 FIFA World Cup in Germany, match results were predicted by an octopus named Paul. As Euro 2016 prepares to kick off Friday in France, scientists are using advanced neural networks to try to figure out which team will win this summer's big soccer tournament. As fans from across the continent begin their journeys toward France this week, predictions among them will be based on passion, patriotism and hope rather than algorithms, artificial intelligence or machine learning. But that's not stopping companies like Microsoft, Yahoo and Blue Yonder from trying to leverage their technology to predict this year's winner. At the World Cup in 2006, Paul the Octopus became a celebrity by accurately predicting the results of every single game involving host country Germany, which went on to win the tournament. Paul's method, though, wasn't what most people would call "scientific."

States across the US are scrambling to figure out how to regulate self-driving cars, wearable technologies that track our health, smart homes that constantly monitor their infrastructure and the rest of the devices emerging from the so-called "internet of things" (IoT). The result is a smattering of incomplete and inconsistent law that could depress the upside of the technology without really addressing its risks. What's most notable about these early regulatory attempts is not that they are varied – that is to be expected. It's that the regulations deal mostly with physical safety, leaving privacy and cybersecurity issues almost wholly unexamined. This seems to be a pattern now, true too of drone regulation, where regulatory bodies have jurisdiction over physical threats, not informational ones.

"Warcraft" just snatched China's all-time midnight presales record from "Furious 7," and it's set to soar to the top of the country's box office as it opens this weekend on more screens than any other film has before. But the video game epic probably won't surpass the car-racing action flick. An estimated one-half of the globe's players of "World of Warcraft" -- the Activision Blizzard video game series on which the movie is based -- live in China, so their enthusiasm was key to setting that presales record. And through late Wednesday, "Warcraft" is off to a sizzling 46 million start in China. Although presales are generally a good proxy for how well a film will do there, the record 391 million haul pulled in by "Furious 7," the highest-grossing Hollywood movie ever in China, should be safe.

The EHang 184 autonomous aerial vehicle is unveiled at the EHang booth at CES International in January in Las Vegas. The drone is large enough to fit a human passenger. The EHang 184 autonomous aerial vehicle is unveiled at the EHang booth at CES International in January in Las Vegas. The drone is large enough to fit a human passenger. The idea: a drone taxi that can transport a single passenger for up to 23 minutes.

It's the death rattles that really get Jack DeGolia, a voice actor in LA who does video game characters. "Blood-curdling screams, choking to death, then that final death rattle," DeGolia recalled. "There's breathing exercises you can do to calm things down, but you need to know your limits and be able to say, 'No'." Most actors won't say no, though: "The risk is, [game makers] will say, 'Fine, we'll get somebody else.'" The union representing actors and performers – Screen Actors Guild?American Federation of Television and Radio Artists, or Sag-Aaftra – has asked the state of California to investigate the video game industry's allegedly brutal vocal recording sessions.