Since the creation of Apache Spark, I/O throughput has increased at a faster pace than processing speed. In a lot of big data applications, the bottleneck is increasingly the CPU. With the release of Apache Spark 2.0 and Project Tungsten, Spark runs a number of control operations close to the metal. At the same time, there has been a surge of interest in using GPUs (the Graphics Processing Units of video cards) for general purpose applications, and a number of frameworks have been proposed to do numerical computations on GPUs. In this talk, we will discuss how to combine Apache Spark with TensorFlow, a new framework from Google that provides building blocks for Machine Learning computations on GPUs.

Next week the boss of one of the UK's largest banking groups will descend on Manchester, to celebrate the opening of the first of Lloyds' new super-size branches. At more than 15,000 sq ft (1,394 sq m) - and at a cost of £3m - it is the bank's most expensive branch ever, and a likely blueprint for other British cities. But while Antonio Horta Osorio will be all smiles, some of his customers may have mixed feelings. Banks may be investing heavily in mega branches in the centre of cities, but those in the suburbs are disappearing fast. In Manchester at least 23 suburban bank branches have closed - or will close - in 2017 alone.

For those in the artificial intelligence community, AI has been around a long time. Originally dating back to the 1950s, AI research has been a fixed part of the landscape since the 1980s becoming a multibillion dollar industry long before the millennium. But for the wider public, AI has only received real attention in the last year or two. During that time, AI has made big news. Google images for AI show a range of anthropomorphized images typically illustrating scare stories of robots supplanting humans in a whole variety of jobs with a clear message echoed recently by Mark Carney: you will soon be redundant.

The ability to automatically recognize a person's behavioral context can contribute to health monitoring, aging care and many other domains. Validating context recognition in-the-wild is crucial to promote practical applications that work in real-life settings. We collected over 300k minutes of sensor data with context labels from 60 subjects. Unlike previous studies, our subjects used their own personal phone, in any way that was convenient to them, and engaged in their routine in their natural environments. Unscripted behavior and unconstrained phone usage resulted in situations that are harder to recognize. We demonstrate how fusion of multi-modal sensors is important for resolving such cases. We present a baseline system, and encourage researchers to use our public dataset to compare methods and improve context recognition in-the-wild.

The issue is now openly on the table. Today marks a significant moment in the life of Estonian digital society: government officials and advisors shine a light on the opportunities of implementing Artificial Intelligence (AI) based solutions as a further upgrade to the already advanced level of the services provided by private and public organizations for the benefit of Estonian residents. Speaking to Marten Kaevats, adviser for digital innovation at the strategy unit of the Government Office of Estonia, it has emerged that experts at Stenbock House have started to discuss feasibility and legal frameworks for the application of Artificial Intelligence technologies to everyday life tasks of Estonian citizens. The spark was represented by the self-driving vehicles experiment carried out during the last months in Tallinn: summer 2017 has seen the first driverless bus being operative on a small route in the city centre of the capital; this time, Estonia has once again the chance to write a new chapter in the history of technological development through an open discussion on how intelligent machines and algorithms are set to change peoples' lives. Self-driving vehicles are just the beginning: "Working only on traffic laws is kind of unreasonable – Kaevats says – because the issue of Artificial Intelligence is much wider and the scope is quite bigger than just dealing with traffic laws".

In the context of things that are dangerous for humans to do, jumping out of an airplane tens of thousands of feet in the air and relying on a combination of fabric and string to keep you from plowing into the ground at a double-digit percentage of the speed of sound seems like it could certainly be one of the more dangerous (even if in reality it isn't). While humans aren't likely to stop doing this anytime soon, autonomous robots may be able to help by mitigating the risk somewhat, offloading hardware and tasks from skydivers to themselves in a chubby little orange package called the Freefall Camera. The Freefall Camera (FFC), which David Alatorre Troncoso and David T. Branson III from the University of Nottingham presented this week at IROS 2017 in Vancouver, is an attempt to autonomously film skydivers in free fall. This is useful for a few reasons: First, it means that skydivers aren't carrying their own cameras (these can get caught on the things that attach you to your parachute, which have a significant amount of importance in avoiding death). And second, it means that one skydiver doesn't need to carry or point a camera if a group of skydivers want to record themselves doing synchronized maneuvers and tricks, which I guess is what happens when just jumping out of an airplane stops being exciting enough by itself.

The cybersecurity industry has always had a fortress mentality: Firewall the perimeter! But that mindset has failed--miserably, as each new headline-generating hack reminds us. Even if you do patch all your software, the way Equifax didn't, or you randomize all your passwords, the way most of us don't, bad actors are going to get past your heavily guarded gate, into your network. And once they do, they're free to go wild. Scott Rosenberg is an editor at Backchannel.

Coffee has been delivered by drone for the first time ever in a densely populated area. The test was part of a project in Zurich to deliver household items such as toothbrushes, deodorant and smartphones to Swiss homes by unmanned aerial devices this autumn. Big firms such as Amazon and Google have spent several years investing in drone delivery research, seeing it as the future of goods distribution. The test was part of a project in Zurich to deliver household items such as toothbrushes, deodorant and smartphones to Swiss homes by unmanned aerial devices this autumn. The experiment come as big firms such as Amazon and Google have spent several years investing in drone delivery research.

As it prepares to open the new Centre for Artificial Intelligence and Robotics, a headquarters in The Hague which will monitor developments in artificial intelligence (AI), the United Nations Interregional Crime and Justice Research Institute (UNICRI) has explained the need for the new center with a warning that robots could destabilize the world. AI, and robots that benefit from it, pose a range of potential threats to humans: from the standard fears of automation and the mass unemployment that follows it, to more dramatic concerns that autonomous killer robots will be deployed by those with nefarious aims -- or that they will be self-directed, for that matter. It will be the task of the UNICRI Centre for Artificial Intelligence and Robotics to second-guess each possible threat. The Guardian reports that UNICRI senior strategic adviser Irakli Beridze said that the team at The Hague will also generate ideas about how AI advances could help achieve UN targets. His point seemed to be that while there are risks associated with developments in AI that needed to be addressed, there is a bigger picture that the center will consider, as the UN's first permanent office focused on AI. "If societies do not adapt quickly enough, this can cause instability," Beridze told the Dutch newspaper de Telegraaf.

Moscow's local authorities are giving the city's 170,000 security cameras a power-up. Moscow's extensive network of security cams have been keeping an eye on the city and recording millions of hours of video since 2012. Processing all that footage proved to be a daunting task, however: police officers quickly realized that it was impossible to look through them all to find culprits to arrest. Authorities believe facial recognition is the answer to that problem. The city will use a facial-recognition technology developed by Russian startup Ntechlab.