OSSENDRECHT, NETHERLANDS – After months of tests, Dutch police will become the world's first force to train and employ an army of eagles, using the centuries-old skill of falconry against the modern-day scourge of unauthorized drones. In their first public demonstration of their unorthodox new weapon, Dutch police on Monday sent out 2-year-old Hunter, a female American bald eagle, and her trainer, Ben. And what better scenario to show off the bird's prowess then a mock-up of a state visit? As the "visiting" head of state, played by a woman police officer, emerged from her car at a police academy in the southern Netherlands to shake hands, a drone suddenly appeared. "Attack, attack," came the cry, while sirens began wailing.

In this form of iterated learning, agents teach each other in sequence: X teaches Y, who then teaches Z, who then teaches... [1-10]. By a classic result of Griffiths and Kalish [3], Quenya will vanish after a finite number of iterations, at which point the agents, assumed to be rational, will be "teaching" each other plain English. In other words, after a while, learners will be taught nothing they don't already know: iterated learning is not self-sustaining. Such findings are hard to validate empirically but variants of it are within the reach of experimental psychology. As early as 1932, in fact, the English psychologist Frederic Bartlett used iterated learning to expose hidden biases among humans. He presented a picture of an owl to a person for given period of time and then asked her to draw it from memory. Her picture was then shown to the next learner for the same amount of time, who then proceeded to draw it back from memory. After 20 iterations of this process, to Bartlett's surprise, what was being drawn was no longer an owl but, quite clearly, a This work was supported in part by NSF grant CCF-1420112.

We formulate and analyze a graphical model selection method for inferring the conditional independence graph of a high-dimensional nonstationary Gaussian random process (time series) from a finite-length observation. The observed process samples are assumed uncorrelated over time and having a time-varying marginal distribution. The selection method is based on testing conditional variances obtained for small subsets of process components. This allows to cope with the high-dimensional regime, where the sample size can be (drastically) smaller than the process dimension. We characterize the required sample size such that the proposed selection method is successful with high probability.

We introduce a new recursive aggregation procedure called Bernstein Online Aggregation (BOA). The exponential weights include an accuracy term and a second order term that is a proxy of the quadratic variation as in Hazan and Kale (2010). This second term stabilizes the procedure that is optimal in different senses. We first obtain optimal regret bounds in the deterministic context. Then, an adaptive version is the first exponential weights algorithm that exhibits a second order bound with excess losses that appears first in Gaillard et al. (2014). The second order bounds in the deterministic context are extended to a general stochastic context using the cumulative predictive risk. Such conversion provides the main result of the paper, an inequality of a novel type comparing the procedure with any deterministic aggregation procedure for an integrated criteria. Then we obtain an observable estimate of the excess of risk of the BOA procedure. To assert the optimality, we consider finally the iid case for strongly convex and Lipschitz continuous losses and we prove that the optimal rate of aggregation of Tsybakov (2003) is achieved. The batch version of the BOA procedure is then the first adaptive explicit algorithm that satisfies an optimal oracle inequality with high probability.

Graph aggregation is the process of computing a single output graph that constitutes a good compromise between several input graphs, each provided by a different source. One needs to perform graph aggregation in a wide variety of situations, e.g., when applying a voting rule (graphs as preference orders), when consolidating conflicting views regarding the relationships between arguments in a debate (graphs as abstract argumentation frameworks), or when computing a consensus between several alternative clusterings of a given dataset (graphs as equivalence relations). In this paper, we introduce a formal framework for graph aggregation grounded in social choice theory. Our focus is on understanding which properties shared by the individual input graphs will transfer to the output graph returned by a given aggregation rule. We consider both common properties of graphs, such as transitivity and reflexivity, and arbitrary properties expressible in certain fragments of modal logic. Our results establish several connections between the types of properties preserved under aggregation and the choice-theoretic axioms satisfied by the rules used. The most important of these results is a powerful impossibility theorem that generalises Arrow's seminal result for the aggregation of preference orders to a large collection of different types of graphs.

Efficient usage of the knowledge provided by the Linked Data community is often hindered by the need for domain experts to formulate the right SPARQL queries to answer questions. For new questions they have to decide which datasets are suitable and in which terminology and modelling style to phrase the SPARQL query. In this work we present an evolutionary algorithm to help with this challenging task. Given a training list of source-target node-pair examples our algorithm can learn patterns (SPARQL queries) from a SPARQL endpoint. The learned patterns can be visualised to form the basis for further investigation, or they can be used to predict target nodes for new source nodes. Amongst others, we apply our algorithm to a dataset of several hundred human associations (such as "circle - square") to find patterns for them in DBpedia. We show the scalability of the algorithm by running it against a SPARQL endpoint loaded with > 7.9 billion triples. Further, we use the resulting SPARQL queries to mimic human associations with a Mean Average Precision (MAP) of 39.9 % and a Recall@10 of 63.9 %.

Not many athletes compete in mind-controlled computer games. Next month, however, more than fifty teams from around the world will meet near Zurich, Switzerland, to demonstrate their skills in manipulating computer characters, going up and down stairs in powered wheelchairs, and racing to pick up objects with their bionic hands. The events, which start on October 8, are part of the world's first-ever Cybathlon, and will bring together the world's best scientists and disabled prosthetic users. But while the Paralympics focuses on outstanding athleticism, the Cybathlon will highlight novel robotic assistive devices that can help people with physical disabilities cope with everyday life. "We are not so much a sports event, and the philosophy is different," Robert Riener, a professor at ETH Zurich University who helped organize the Cybathlon told GeekWire.

Why we urgently need to measure AI's societal impacts How will artificial intelligence systems change the way we live? This is a tough question: on one hand, AI tools are producing compelling advances in complex tasks, with dramatic improvements in energy consumption, audio processing, and leukemia detection. There is extraordinary potential to do much more in the future. On the other hand, AI systems are already making problematic judgements that are producing significant social, cultural, and economic impacts in people's everyday lives. AI and decision-support systems are embedded in a wide array of social institutions, from influencing who is released from jail to shaping the news we see.

Grand Theft Auto may not be the first place you'd go to learn better driving skills, but researchers are now using this fictional world to train self-driving cars. A team has discovered that machine learning can extract maneuver and landscape data much faster in these virtual settings than with traditional methods. Using computer vision algorithms, the team labeled thousands of images in just a few hours, which will be used to teach autonomous cars to recognize different objects. Cars use object identification data to'learn' how to recognize objects, such as pedestrians or cyclists. Gathering this information is usually conducted by researchers, who comb through scenes from real life footage and draw borders around objects by hand.

As the enterprise becomes more steeped in machine learning, cognitive computing and other forms of autonomous infrastructure, the issue of artificial intelligence ethics keeps coming up. Science fiction abounds with tales of benign computer systems that suddenly conclude people are simply inadequate and set about destroying humanity. As amusing as it is, though, how realistic is it that artificial intelligence (AI) could someday pose a hazard due to its development as it continues to evolve? Are there any reasons (and surefire means) to keep it in check? According to WebVisions, there are two ways to program ethics into machines: Hard code them into their operating systems or establish guidelines that allow these machines to reach ethical conclusions on their own.