Sir Anthony Seldon (pictured) warned that the technology could manipulate young people into doing'unspeakable' things, by impersonating teachers or celebrities Online predators could use artificial intelligence to manipulate children into'unspeakable' actions by posing as teachers or celebrities, a professor has warned. Sir Anthony Seldon said people should'wake up and smell the silicone' about the risk AI poses to children. Speaking at the launch of the UK's first Institute for Ethical AI in Education (IEAIED), warned schools were not teaching'critical thinking' to help pupils fend off dangerous technology. Sir Anthony, vice-chancellor of the University of Buckingham, said: 'These machines can impersonate teachers, they can either impersonate people who they don't know but look plausible, they can impersonate public figures. 'Or they can impersonate teachers – the headteacher of the school – and insinuate and manipulate children into thinking that this is the real teacher telling them to do things.

Last week we wrapped up a highly successful GPU Technology Conference (GTC) Europe in Munich! GTC is NVIDIA's international conference series, bringing together the top minds in deep learning, analytics, and of course GPUs for sessions, workshops, keynotes, and more. This was the place to be for any and all European organizations interested in leveraging the power of the GPU. As the Kinetica engine runs on GPUs, there's no better place for us to share our solutions for advanced analytics and deep learning. This year we noticed a significant increase in the number of organizations that understand the challenges of the Extreme Data Economy.

Europe has a fascinating array of data science companies transforming the world of business. Here are the ones to watch. When it comes to areas such as data science, artificial intelligence (AI) and other avenues of deep tech, Europe is killing it. Last November at Slush in Helsinki, Tom Wehmeier, partner and research director at venture capital firm Atomico, revealed how European start-ups are now producing more software engineers than their US counterparts, with 5.5m compared with 4.4m in the US. He noted that European tech firms such as Zalando and Spotify are absorbing the battle for talent by building their companies in a distributed way across Europe.

A British-built spacecraft is about to blast off to Mercury. The BepiColombo ship will fly five billion miles to the planet that is closest to our sun. The scientists behind it hope that it can help solve some of the mysteries that hang over the distant world. It might, for instance, contain water ice, hidden in shadows on its otherwise blazing hot surface, scientists speculate – a discovery that could help shed light on the possibility of alien life in our own solar system. It will also look to explain why the planet has such as huge iron core at its centre, and unlock the history of the vast volcanic vents that shook its surface for billions of years.

Reading medieval literature, it's hard not to be impressed with how much the characters get done--as when we read about King Harold doing battle in one of the Sagas of the Icelanders, written in about 1230. The first sentence bristles with purposeful action: "King Harold proclaimed a general levy, and gathered a fleet, summoning his forces far and wide through the land." By the end of the third paragraph, the king has launched his fleet against a rebel army, fought numerous battles involving "much slaughter in either host," bound up the wounds of his men, dispensed rewards to the loyal, and "was supreme over all Norway." What the saga doesn't tell us is how Harold felt about any of this, whether his drive to conquer was fueled by a tyrannical father's barely concealed contempt, or whether his legacy ultimately surpassed or fell short of his deepest hopes. In his short story "Forever Overhead," the 13-year-old protagonist takes 12 pages to walk across the deck of a public swimming pool, wait in line at the high diving board, climb the ladder, and prepare to jump.

Business leaders put too much faith in using artificial intelligence (AI) to solve their cyber security problems, and should instead focus on educating users on cyber hygiene and managing risks, according to a cyber security expert. Security technologist Bruce Schneier's insights and warnings around the regulation of IoT security and forensic cyber psychologist Mary Aiken's comments around the tensions between encryption and state security were the top highlights of the keynote presentations at Infosecurity Europe 2017 in London. You forgot to provide an Email Address. This email address doesn't appear to be valid. This email address is already registered.

Drone technology has advanced markedly in the last few years, with improved stability and handling making them easier than ever to fly. But whether through mechanical malfunction or sheer pilot incompetence, there will always be occasions when we're left watching helplessly as our crippled quadcopter plummets from the sky as if it was never meant to be up there in the first place. One solution is to stick a parachute on the drone that automatically activates when it detects problems. Such a system would not only save the drone from breaking into multiple pieces when it hits terra firma, but also reduce the risk of injury if the machine lands on someone's head on the way down. Among a growing number of such offerings is one from Austria-based Drone Rescue, which has been working on incorporating parachutes into drones for a while now.

In this paper, we provide an overview of first-order and second-order variants of the gradient descent methods commonly used in machine learning. We propose a general framework in which 6 of these methods can be interpreted as different instances of the same approach. These methods are the vanilla gradient descent, the classical and generalized Gauss-Newton methods, the natural gradient descent method, the gradient covariance matrix approach, and Newton's method. Besides interpreting these methods within a single framework, we explain their specificities and show under which conditions some of them coincide. Machine learning generally amounts to solving an optimization problem where a loss function has to be minimized.

New communication standards need to deal with machine-to-machine communications, in which users may start or stop transmitting at any time in an asynchronous manner. Thus, the number of users is an unknown and time-varying parameter that needs to be accurately estimated in order to properly recover the symbols transmitted by all users in the system. In this paper, we address the problem of joint channel parameter and data estimation in a multiuser communication channel in which the number of transmitters is not known. For that purpose, we develop the infinite factorial finite state machine model, a Bayesian nonparametric model based on the Markov Indian buffet that allows for an unbounded number of transmitters with arbitrary channel length. We propose an inference algorithm that makes use of slice sampling and particle Gibbs with ancestor sampling. Our approach is fully blind as it does not require a prior channel estimation step, prior knowledge of the number of transmitters, or any signaling information. Our experimental results, loosely based on the LTE random access channel, show that the proposed approach can effectively recover the data-generating process for a wide range of scenarios, with varying number of transmitters, number of receivers, constellation order, channel length, and signal-to-noise ratio.

Stochastic variational inference is an established way to carry out approximate Bayesian inference for deep models. While there have been effective proposals for good initializations for loss minimization in deep learning, far less attention has been devoted to the issue of initialization of stochastic variational inference. We address this by proposing a novel layer-wise initialization strategy based on Bayesian linear models. The proposed method is extensively validated on regression and classification tasks, including Bayesian DeepNets and ConvNets, showing faster convergence compared to alternatives inspired by the literature on initializations for loss minimization.