With a focus on chips and artificial intelligence, U.K.-based Graphcore can now be considered one of Europe's hottest startups. Today, the company announced it has raised a $50 million round of funding led by Silicon Valley's Sequoia Capital, a firm not known for investing much in Europe. This follows the $60 million that Graphcore had already raised over the last 18 months. In a blog post, Graphcore cofounder Nigel Toon wrote that the company's partnership with Sequoia is an indication that it intends to remain independent as it seeks to compete in the surging AI chip market. "So over the last few weeks, Graphcore and Sequoia Capital have worked together on a scale-up business plan and on a funding plan which will allow us to grow more quickly and to support our prospective customers more deeply as we bring products to market," Toon wrote.

The race to conquer the driverless car market has stepped up a gear, with the first ever tests of an autonomous vehicle built in Britain on the country's public roads. Jaguar Land Rover is leading the pack with its'major landmark' trial, which aims to help vehicles react in a similar way to people. The pilot project is part of a government-backed bid to encourage more widespread use of automated cars by 2020. The race to conquer the driverless car market has stepped up a gear, with the first ever tests of an autonomous vehicle built in Britain on the country's public roads. The UK Autodrive project is the UK's largest trial of connected and autonomous vehicle technology.

For as long as artificial intelligence and machine learning tools have been moving into the workforce, there have been rumblings of robots taking over the work of people, and the impact that could have on their career prospects. However, new studies undertaken by global professional services brand Genpact of 5,000 respondents in the United Kingdom, United States, and Australia, shows that the level of concern among the workers themselves is not very high. Roughly twenty percent of those surveyed in the UK felt that their jobs were threatened by AI, with only six percent feeling this strongly. But, although they did not feel overly cautious about their own prospects, they saw the potential disadvantages for the next generation of workers, with over fifty percent responding there was a threat to their children's careers, and over eighty percent stating that new skills will be needed for those workers in order to succeed in an AI advanced environment. The reason for this caution can be found in the training, or lack thereof, in the use of AI.

Slowly, the UK government is realising its dream of making the nation a self-driving research hub. UK Autodrive, a publicly funded consortium that includes Jaguar Land Rover, Ford and TATA Motors, has announced a new set of trials in Coventry today. They will focus on self-driving cars and vehicles that can instantly share information with other motorists and city infrastructure. Researchers will be testing a signal, for instance, that can be sent out by the emergency services -- ambulances, fire trucks and police cars -- to nearby drivers, advising them when and where to move aside. Other test features include a warning signal for intersections deemed too unsafe to cross, in-car information about accidents and traffic jams (negating the need for signs on bridges) and an alert system when a driver in front suddenly hits the brakes (the idea being that this can be hard to spot in rain and fog). UK Autodrive is also looking at connected traffic lights that could help self-driving vehicles optimise their speed, avoid red lights and reduce road congestion.

Singapore enterprises are slower than their counterparts in Asia-Pacific, North America and Europe in adopting machine leaning, despite seeing the importance of automation in handling menial, repetitive tasks, according to a study released today. Only 32 per cent of Singapore chief information officers (CIOs) interviewed say their organisations are already using machine learning in some part of their businesses. This compares with 59 per cent in Australia and New Zealand (49 per cent), in a survey commissioned by enterprise software vendor ServiceNow. In telephone interviews this year, it had asked 500 CIOs in 10 countries for their views on machine learning and automation. They included the United States, France and Britain.

"Anyone who wants to be 100% sure, will be 100% late." I heard this sentence recently at the IoT Forum in Munich. It is a very good summary of the feeling that is emerging in the debate on Industry 4.0. The buzzword battles of the past 48 months seem to be just that; buzzword battles. Stakeholders seem to be stuck waiting for the proposition with guarantees.

We propose a test of independence of two multivariate random vectors, given a sample from the underlying population. Our approach, which we call MINT, is based on the estimation of mutual information, whose decomposition into joint and marginal entropies facilitates the use of recently-developed efficient entropy estimators derived from nearest neighbour distances. The proposed critical values, which may be obtained from simulation (in the case where one marginal is known) or resampling, guarantee that the test has nominal size, and we provide local power analyses, uniformly over classes of densities whose mutual information satisfies a lower bound. Our ideas may be extended to provide a new goodness-of-fit tests of normal linear models based on assessing the independence of our vector of covariates and an appropriately-defined notion of an error vector. The theory is supported by numerical studies on both simulated and real data.

We use deep feedforward artificial neural networks to approximate solutions of partial differential equations of advection and diffusion type in complex geometries. We derive analytical expressions of the gradients of the cost function with respect to the network parameters, as well as the gradient of the network itself with respect to the input, for arbitrarily deep networks. The method is based on an ansatz for the solution, which requires nothing but feedforward neural networks, and an unconstrained gradient based optimization method such as gradient descent or quasi-Newton methods. We provide detailed examples on how to use deep feedforward neural networks as a basis for further work on deep neural network approximations to partial differential equations. We highlight the benefits of deep compared to shallow neural networks and other convergence enhancing techniques.

Achieving superhuman playing level by AlphaGo corroborated the capabilities of convolutional neural architectures (CNNs) for capturing complex spatial patterns. This result was to a great extent due to several analogies between Go board states and 2D images CNNs have been designed for, in particular translational invariance and a relatively large board. In this paper, we verify whether CNN-based move predictors prove effective for Othello, a game with significantly different characteristics, including a much smaller board size and complete lack of translational invariance. We compare several CNN architectures and board encodings, augment them with state-of-the-art extensions, train on an extensive database of experts' moves, and examine them with respect to move prediction accuracy and playing strength. The empirical evaluation confirms high capabilities of neural move predictors and suggests a strong correlation between prediction accuracy and playing strength. The best CNNs not only surpass all other 1-ply Othello players proposed to date but defeat (2-ply) Edax, the best open-source Othello player.

We present a weakly-supervised approach to segmenting proposed drivable paths in images with the goal of autonomous driving in complex urban environments. Using recorded routes from a data collection vehicle, our proposed method generates vast quantities of labelled images containing proposed paths and obstacles without requiring manual annotation, which we then use to train a deep semantic segmentation network. With the trained network we can segment proposed paths and obstacles at run-time using a vehicle equipped with only a monocular camera without relying on explicit modelling of road or lane markings. We evaluate our method on the large-scale KITTI and Oxford RobotCar datasets and demonstrate reliable path proposal and obstacle segmentation in a wide variety of environments under a range of lighting, weather and traffic conditions. We illustrate how the method can generalise to multiple path proposals at intersections and outline plans to incorporate the system into a framework for autonomous urban driving.