Olli will be exclusive to DC these next few months, but Miami and Las Vegas will get their own in late 2016. Local Motors is also in talks to test the bus in cities outside the US, including Berlin, Copenhagen and Canberra. It's unclear if anyone can get the chance to ride one, since these are merely trial runs, but you can ask local authorities if the EV makes its way to your city. If and when the time comes that driverless public vehicles can legally shuttle passengers, you'll be able to summon an Olli through an app, just like Uber. And if Local Motors' plans pan out, a lot of people around the globe will be using that app: Company co-founder John Rogers envisions building hundreds of micro-factories all over the world that can 3D print an Olli within 10 hours and assemble it one.

A robotic employee will be deployed instead of human council workers to answer customer queries. IPsoft said Amelia, its technology platform, will be deployed to work for Enfield Council in North London. Capable of analyzing natural language, she understands context, applies logic, learns, resolves problems and even senses emotions. IPsoft said Amelia will be deployed to work for Enfield Council in North London. Capable of analyzing natural language, she understands context and even senses emotions.

This is a real laser gun that's marketed to real militaries in our real reality. Drones are smaller, slower, cheaper, and lower-flying targets than most airplanes, which makes them a weird threat to modern militaries. Even when just scouting, and especially if outfitted with explosives, the unmanned aerial vehicles are dangerous enough to warrant their destruction, but cheap enough that it doesn't make sense to use a missile. Enter the laser cannons, like this one on display at the Eurosatory 2016 Land and Airland Defence and Security tradeshow. Made by Germany's Rheinmetall, the Oerlikon Skyshield High Energy Laser is part of a larger system of sensors and weapons.

How do scientists decide what research to do? One would like to think that they take a suitably scientific approach to this question by thinking about important problems that need to be solved, and asking which of these problems could be solved given the time and money available. But are research projects actually proposed and funded in this way, or are there other forces at work? Particle physicists and astronomers realized decades ago that they needed to take a coordinated approach to planning so that they had accelerators and telescopes to work on. This "big science" approach involved agreeing on the long-term scientific goals in a given field and then getting the relevant funding agencies in different countries on board. This approach has been remarkably successful, as demonstrated by the recent detections of the Higgs boson and gravitational waves.

In addition to conducting pure research in artificial intelligence and machine learning (ML), the group will also develop new tools and products that make use of the technology. The company noted that it already offers several services to consumers that are based on machine learning technology, such as Google Translate, Photo Search, and Smart Reply for Inbox. The new research center is part of a broader initiative at Google to advance the field of machine learning. Researchers working at the company's existing engineering offices in Zurich have already made major contributions to the field, such as developing the conversation engine that powers the Google Assistant in the Allo smart messaging app, and developing the engine that powers Knowledge Graph. The Zurich office is already the company's largest engineering office outside of the US.

The importance of algorithms in our lives today cannot be overstated. They are used virtually everywhere, from financial institutions to dating sites. But some algorithms shape and control our world more than others -- and these ten are the most significant. Just a quick refresher before we get started. Though there's no formal definition, computer scientists describe algorithms as a set of rules that define a sequence of operations.

Retailers and technology firms are experimenting with chatbots, powered by a combination of machine learning, natural language processing, and live operators, to provide customer service, sales support, and other commerce-related functions. Chris Messina of Uber recently coined the term "conversational commerce" to describe this movement, which he defines as: The net result is that you and I will be talking to brands and companies over Facebook Messenger, WhatsApp, Telegram, Slack, and elsewhere before year's end, and will find it normal. While messaging and voice interfaces are central components, they fit into a larger picture of increasing infusion of technology into our daily lives, which in turn is unlocking new potential for brand-to-consumer interaction. The fact is, technology overall is becoming more deeply woven into our lives, and the entire ecosystem is enjoying tighter cohesion through the increasing availability and sophistication of APIs. Smart companies are finding new and innovative touch points with consumers that are contextual, relevant, highly personal, and, yes, conversational.

We combine Riemannian geometry with the mean field theory of high dimensional chaos to study the nature of signal propagation in generic, deep neural networks with random weights. Our results reveal an order-to-chaos expressivity phase transition, with networks in the chaotic phase computing nonlinear functions whose global curvature grows exponentially with depth but not width. We prove this generic class of deep random functions cannot be efficiently computed by any shallow network, going beyond prior work restricted to the analysis of single functions. Moreover, we formalize and quantitatively demonstrate the long conjectured idea that deep networks can disentangle highly curved manifolds in input space into flat manifolds in hidden space. Our theoretical analysis of the expressive power of deep networks broadly applies to arbitrary nonlinearities, and provides a quantitative underpinning for previously abstract notions about the geometry of deep functions.

Improving the interpretability of brain decoding approaches is of primary interest in many neuroimaging studies. Despite extensive studies of this type, at present, there is no formal definition for interpretability of brain decoding models. As a consequence, there is no quantitative measure for evaluating the interpretability of different brain decoding methods. In this paper, we present a simple definition for interpretability of linear brain decoding models. Then, we propose to combine the interpretability and the performance of the brain decoding into a new multi-objective criterion for model selection. Our preliminary results on the toy data show that optimizing the hyper-parameters of the regularized linear classifier based on the proposed criterion results in more informative linear models. The presented definition provides the theoretical background for quantitative evaluation of interpretability in linear brain decoding.

In the standard setting of approachability there are two players and a target set. The players play repeatedly a known vector-valued game where the first player wants to have the average vector-valued payoff converge to the target set which the other player tries to exclude it from this set. We revisit this setting in the spirit of online learning and do not assume that the first player knows the game structure: she receives an arbitrary vector-valued reward vector at every round. She wishes to approach the smallest ("best") possible set given the observed average payoffs in hindsight. This extension of the standard setting has implications even when the original target set is not approachable and when it is not obvious which expansion of it should be approached instead. We show that it is impossible, in general, to approach the best target set in hindsight and propose achievable though ambitious alternative goals. We further propose a concrete strategy to approach these goals. Our method does not require projection onto a target set and amounts to switching between scalar regret minimization algorithms that are performed in episodes. Applications to global cost minimization and to approachability under sample path constraints are considered.