Getting dental work done is generally very expensive and you have to basically estimate by looking at casts and molds as to how your teeth will look post-treatment. A Swiss startup is working on changing this paradigm. Kapanu has created a virtual mirror which will show you your post-treatment teeth using augmented reality (AR) and 3D imaging. The company has created an AR engine in collaboration with Disney Research, which will be used for medical purposes, the first one being dentistry. The company showcased its tech at the International Dental Show in Cologne, Germany, last year and claims that it received "overwhelming feedback" from dentists, dental hygienists and dental technicians.

This week sees thousands of children throughout the country wake up and realise with stark horror that the summer holidays are over and school beckons. Most adults can remember the sudden system shock of these mornings; the alarm going off unreasonably early, the shivering cold of the bathroom, the family gathered in stony silence around the table, munching forlornly on soggy toast. Games such as Resident Evil or Silent Hill have conjured few horrors that compare with entering a new classroom and meeting an unfamiliar teacher who may or may not prove to be an authoritarian sociopath. This sense of fear and loathing was perhaps why, when my dad used to get home from work and find me watching Grange Hill, he would always tut and say'haven't you had enough of school?'. But of course, for several generations of kids in the UK, Grange Hill was our way of confronting and processing the horrors of secondary education.

Scientists are seeking the public's assistance in rescuing a unique set of weather records gathered at the summit of the UK's highest mountain. From 1883 to 1904, meteorologists were stationed atop Ben Nevis, logging temperature, precipitation, wind and other data around the clock. Their measurements are held in five big volumes that now need to be digitised to be useful to modern researchers. The public can help with the conversion at the www.weatherrescue.org It will involve copying tables into a database.

As automation and artificial intelligence technologies improve, many people worry about the future of work. As automation and artificial intelligence technologies improve, many people worry about the future of work. If millions of human workers no longer have jobs, the worriers ask, what will people do, how will they provide for themselves and their families, and what changes might occur (or be needed) in order for society to adjust? Many economists say there is no need to worry. They point to how past major transformations in work tasks and labor markets – specifically the Industrial Revolution during the 18th and 19th centuries – did not lead to major social upheaval or widespread suffering.

Do you hear the clunking sounds? Those are robots marching to take your job and put you on the brink of grim unemployment survival. Before you start frantically examining your job description, let's figure out what's going on. Even if you are the one who bothers to read the copy that follows the headlines, succumbing to alarmist stories is not hard to do, even for Huffington Post readers. Reporters often fail to explain exactly what stands behind the numbers. The most quoted study that estimates jobs susceptible to automation is the work by Carl Frey and Michael Osborne (so-called FO) The Future of Employment published in 2013 by Oxford University. This scientific research conducted four years ago still serves as the foundation of many predictions to render them more academically credible. And yes, it's them who estimated 47 percent.

Decentralised analytics company Cindicator has undertaken a pilot project with the Moscow Stock Exchange, which showed the platform was able to drive an estimated 47% yield per annum for an experimental investment portfolio. Cindictor, which recently raised $500,000 (£386,000) in seed investment, forecasts financial case outcomes combining data from 15,000 non-professional analysts and AI mechanics, to provide hedge funds and institutional investors with precise forecasts. The pilot project saw a pool of 863 independent non-professional analysts' predicted price points for four futures daily. Based on answers to 56 questions, the platform powered around 100 deals, more than 80% of which turned out profitable. In 15 days, a model portfolio increased by 2.81% in value, which equals a 47% yield per annum.

It seems window shopping on Tinder is becoming as common as scrolling through Instagram, even among those who are off the market. According to new research, 35 per cent of British people still use dating apps when in a relationship. Despite the shocking stastistic, one in five of the 2,100 people surveyed by specialist law firm Slater and Gordon said this is just part and parcel of the modern day dating life and they'wouldn't mind' if their partner used a dating app. The younger generations seemed even more accepting with two thirds of 16-24 years old admitting they would happily forgive their other-half if they found them checking out other'options' online. Just under half of the men (46 per cent) who were surveyed admitted to using the modern dating tool while in a relationship.

Networks capture pairwise interactions between entities and are frequently used in applications such as social networks, food networks, and protein interaction networks, to name a few. Communities, cohesive groups of nodes, often form in these applications, and identifying them gives insight into the overall organization of the network. One common quality function used to identify community structure is modularity. In Hu et al. [SIAM J. App. Math., 73(6), 2013], it was shown that modularity optimization is equivalent to minimizing a particular nonconvex total variation (TV) based functional over a discrete domain. They solve this problem---assuming the number of communities is known---using a Merriman, Bence, Osher (MBO) scheme. We show that modularity optimization is equivalent to minimizing a convex TV-based functional over a discrete domain---again, assuming the number of communities is known. Furthermore, we show that modularity has no convex relaxation satisfying certain natural conditions. Despite this, we partially relax the discrete constraint using a Ginzburg Landau functional, yielding an optimization problem that is more nearly convex. We then derive an MBO algorithm with fewer parameters than in Hu et al. and which is 7 times faster at solving the associated diffusion equation due to the fact that the underlying discretization is unconditionally stable. Our numerical tests include a hyperspectral video whose associated graph has 29 million edges, which is roughly 37 times larger than was handled in the paper of Hu et al.

We present a practical way of introducing convolutional structure into Gaussian processes, making them more suited to high-dimensional inputs like images. The main contribution of our work is the construction of an inter-domain inducing point approximation that is well-tailored to the convolutional kernel. This allows us to gain the generalisation benefit of a convolutional kernel, together with fast but accurate posterior inference. We investigate several variations of the convolutional kernel, and apply it to MNIST and CIFAR-10, which have both been known to be challenging for Gaussian processes. We also show how the marginal likelihood can be used to find an optimal weighting between convolutional and RBF kernels to further improve performance. We hope that this illustration of the usefulness of a marginal likelihood will help automate discovering architectures in larger models.

Célia Châtel Aix-Marseille University, France Pierre Geurts University of Liège, Belgium Dealing with datasets of very high dimension is a major challenge in machine learning. In this paper, we consider the problem of feature selection in applications where the memory is not large enough to contain all features. In this setting, we propose a novel tree-based feature selection approach that builds a sequence of randomized trees on small subsamples of variables mixing both variables already identified as relevant by previous models and variables randomly selected among the other variables. As our main contribution, we provide an in-depth theoretical analysis of this method in infinite sample setting. In particular, we study its soundness with respect to common definitions of feature relevance and its convergence speed under various variable dependance scenarios. We also provide some preliminary empirical results highlighting the potential of the approach.