Participants run ahead of Puerto de San Lorenzo's fighting bulls during the third bull run of the San Fermin festival in Pamplona, northern Spain. Each day at 8:00 am hundreds of people race with six bulls, charging along a winding, 848.6-metre (more than half a mile) course through narrow streets to the city's bull ring, where the animals are killed in a bullfight or corrida, during this festival, immortalised in Ernest Hemingway's 1926 novel "The Sun Also Rises" and dating back to medieval times and also featuring religious processions, folk dancing, concerts and round-the-clock drinking. Iraqi women, who fled the fighting between government forces and Islamic State (IS) group jihadists in the Old City of Mosul, cry as they stand in the city's western industrial district awaiting to be relocated

Samsung's virtual assistant Bixby has added voice commands for UK users for the first time. The artificial intelligence-powered assistant first appeared on Samsung's Galaxy S8 and S8 smartphones earlier this year and enables users to ask questions and quickly access different parts of their smartphone using command prompts. Until now, voice commands had only been available in the US and South Korea, with only text and camera-based feature accessible to users outside these countries. But the UK is now one of more than 200 countries included in the expansion, which can be triggered by saying "Hi, Bixby". Similar to other smart assistant's including Apple's Siri, Google Assistant and Amazon's Alexa, Bixby understands and responds to voice commands on scheduling, news updates and weather reports, as well as launching apps on a user's smartphone.

Last week, Qualcomm announced that it had acquired Netherlands-based machine learning startup Scyfer for an undisclosed amount. The company also laid out its vision for artificial intelligence, indicating that it would double down on a device-focused AI implementation. Below, we provide a quick run-down of what Qualcomm has been doing in the AI space and what Scyfer could bring to the table. Trefis has a $64 price estimate for Qualcomm, which is about 20% ahead of the current market price. While Qualcomm has been working on AI for about a decade, the company's more recent efforts have centered around deploying artificial intelligence technology at the device level – in smartphones and cars.

About five or six years ago, one of Karl Ricanek's students showed him a video on YouTube. It was a time lapse of a person undergoing hormone replacement therapy, or HRT, in order to transition genders. "At the time, we were working on facial recognition," Ricanek, a professor of computer science at the University of North Carolina at Wilmington, tells The Verge. He says he and his students were always trying to find ways to break the systems they worked on, and that this video seemed like a particularly tricky challenge. "We were like, 'Wow there's no way the current technology could recognize this person [after they transitioned].'"

An increasingly popular concern is that robots will eat up labour's share of income at an accelerating rate, leaving ordinary workers impoverished and unemployed. A common dinner conversation topic in Silicon Valley is universal basic income, and the typical argument advanced for UBI is that we are destined to indefinitely continue losing jobs faster than we replace them. Variants on this theme have circulated since the dawn of the Industrial Revolution. Improvements in farming technology have been greeted with skepticism since ancient times for these reasons. Mechanical contraptions for sewing and other tasks were decried as potentially ruinous to workers in Elizabethan England.

We consider a variant of online convex optimization in which both the instances (input vectors) and the comparator (weight vector) are unconstrained. We exploit a natural scale invariance symmetry in our unconstrained setting: the predictions of the optimal comparator are invariant under any linear transformation of the instances. Our goal is to design online algorithms which also enjoy this property, i.e. are scale-invariant. We start with the case of coordinate-wise invariance, in which the individual coordinates (features) can be arbitrarily rescaled. We give an algorithm, which achieves essentially optimal regret bound in this setup, expressed by means of a coordinate-wise scale-invariant norm of the comparator. We then study general invariance with respect to arbitrary linear transformations. We first give a negative result, showing that no algorithm can achieve a meaningful bound in terms of scale-invariant norm of the comparator in the worst case. Next, we compliment this result with a positive one, providing an algorithm which "almost" achieves the desired bound, incurring only a logarithmic overhead in terms of the norm of the instances. Keywords: Online learning, online convex optimization, scale invariance, unconstrained online learning, linear classification, regret bound.

Deep neural networks have been widely adopted in recent years, exhibiting impressive performances in several application domains. It has however been shown that they can be fooled by adversarial examples, i.e., images altered by a barely-perceivable adversarial noise, carefully crafted to mislead classification. In this work, we aim to evaluate the extent to which robot-vision systems embodying deep-learning algorithms are vulnerable to adversarial examples, and propose a computationally efficient countermeasure to mitigate this threat, based on rejecting classification of anomalous inputs. We then provide a clearer understanding of the safety properties of deep networks through an intuitive empirical analysis, showing that the mapping learned by such networks essentially violates the smoothness assumption of learning algorithms. We finally discuss the main limitations of this work, including the creation of real-world adversarial examples, and sketch promising research directions.

A network is composed of a set of nodes, or vertices, with connections between them. Network data arise in a wide range of fields, and include social networks, collaboration networks, communication networks, biological networks, food webs and are a useful way of representing interactions between sets of objects. Of particular importance is the elaboration of random graph models, which can capture the salient properties of real-world graphs. Following the seminal work of Erd os and R enyi (1959), various network models have been proposed; see the overviews of Newman (2003b, 2009), Kolaczyk (2009), Bollob as (2001), Goldenberg et al. (2010), Fienberg (2012) or Jacobs and Clauset (2014). In particular, a large body of the literature has concentrated on models that can capture some modular or community structure within the network. The first statistical network model in this line of research is the popular stochastic block-model (Holland et al., 1983; Snijders and Nowicki, 1997; Nowicki and Snijders, 2001). The stochastic block-model assumes that each node belongs to one ofp latent communities, and the probability of connection between two nodes is given by ap p connectivity matrix. This model has been extended in various directions, by introducing degree-correction parameters (Karrer and Newman, 2011), by allowing the number of communities to grow with the size of the network (Kemp et al., 2006), or by considering overlapping communities (Airoldi et al., 2008; Miller et al., 2009; Latouche et al., 2011; Palla et al., 2012; Yang and Leskovec, 2013). Stochastic block-models and their extensions have shown to offer a very flexible modeling framework, with interpretable parameters, and have been successfully used for the analysis of numerous real-world networks.

But no one has actually done the tests that could reveal what would happen and inform safety. That's strange, given the increasing risk of such an incident. We need to know if it could cause an aircraft engine to explode in what is known as an "uncontained failure", with hot, fast-spinning engine parts being shed in all directions, potentially piercing wings, fuel tanks and even the cabin.

Willis Towers Watson's insurance consulting business works with major insurers and other financial services companies around the world on issues that range from economic capital and regulatory change to the intricacies of hedging and other forms of asset-liability management. You will deliver high quality work for our broad set of UK clients, working on projects including behavioural modelling, price optimisation, financial risk modelling and big data analytics. You will join a team that works with a huge variety of leading financial institutions across the world, including almost all of the top UK insurers as well as intermediaries, data and software houses and other financial and non-financial organisations. You will have immediate exposure to real client assignments which will draw on your imagination and creativity as well as your ability to analyse data, draw insight and present results. These experiences will help build on your technical knowledge as well as deliver high profile projects to household name clients.