A common way to train neural networks is the Backpropagation. This algorithm includes a gradient descent method, which needs an adaptive step size. In the area of neural networks, the ADAM-Optimizer is one of the most popular adaptive step size methods. It was invented in \cite{Kingma.2015} by Kingma and Ba. The $5865$ citations in only three years shows additionally the importance of the given paper. We discovered that the given convergence proof of the optimizer contains some mistakes, so that the proof will be wrong. In this paper we give an improvement to the convergence proof of the ADAM-Optimizer.

This work introduces CLBlast, an open-source BLAS library providing optimized OpenCL routines to accelerate dense linear algebra for a wide variety of devices. It is targeted at machine learning and HPC applications and thus provides a fast matrix-multiplication routine (GEMM) to accelerate the core of many applications (e.g. deep learning, iterative solvers, astrophysics, computational fluid dynamics, quantum chemistry). CLBlast has five main advantages over other OpenCL BLAS libraries: 1) it is optimized for and tested on a large variety of OpenCL devices including less commonly used devices such as embedded and low-power GPUs, 2) it can be explicitly tuned for specific problem-sizes on specific hardware platforms, 3) it can perform operations in half-precision floating-point FP16 saving bandwidth, time and energy, 4) it has an optional CUDA back-end, 5) and it can combine multiple operations in a single batched routine, accelerating smaller problems significantly. This paper describes the library and demonstrates the advantages of CLBlast experimentally for different use-cases on a wide variety of OpenCL hardware.

We consider the problem of inductive matrix completion under the assumption that many features are non-informative, which leads to row- and column-sparse structure of coefficient matrix. Under the additional assumption on the low rank of coefficient matrix we propose the matrix factorization framework with group-lasso regularization on parameter matrices. We suggest efficient optimization algorithm for the solution of the obtained problem. From theoretical point of view, we prove the oracle generalization bound on the expected error of matrix completion. Corresponding sample complexity bounds show the benefits of the proposed approach compared to competitors in the sparse problems. The experiments on synthetic and real-world datasets show the state-of-the-art efficiency of the proposed method.

Rick Dickinson, the designer of Sinclair computers, has died in the US while receiving treatment for cancer. The British designer, thought to be in his 60s, worked in-house for Sinclair Research and oversaw the creation of its home computers in the 1980s. He was responsible for the boxy look of the ZX80 and ZX81 and the Bauhaus-inspired appearance of the Spectrum. Mr Dickinson also helped to develop the technologies for the UK company's touch-sensitive and rubber keyboards. He was recently linked to a crowd-funded project by Retro Computers to turn the Spectrum into a handheld computer.

Who will win the race to adopt artificial intelligence for cyber warfare--the defenders of vulnerable corporate networks or the cyber criminals constantly inventing new ways to attack them? The promise--or unrealistic hope--that AI will "transform the world," has given rise to a number of significant races. Most prominent is the race among nations for AI superiority, primarily the U.S. and China, with several European countries (e.g., France) and the European Union attempting to position themselves not too far behind. This reminds one of the nuclear arms-race and the use of the same technology for both beneficial and destructive purposes. In a recent paper warning of the potential of AI to "upend the foundations of nuclear deterrence," RAND researchers wrote: "The dual-use nature of many AI algorithms will mean AI research focused on one sector of society can be rapidly modified for use in the security sector as well."

US and China are pulling away in the race to build AI. A group of renowned artificial intelligence (AI) scientists have called for a new multinational AI hub to be built in Europe in a bid to help the continent compete with other parts of the world. AI is set to have a profound impact on Europe and the rest of the world in the coming decades and many believe the impact will be larger than that of the industrial revolution. In an open letter, the scientists wrote that "Europe is not keeping up" with North America and China and called for a new European Lab for Learning & Intelligent Systems, abbreviated as ELLIS. The group argues there are clusters of AI excellence within labs scattered across Europe "that play in the international top league" but "virtually all of the top people in those places are continuously being pursued for recruitment by US companies."

The applications of artificial intelligence to enhance humans' life open a vast array of opportunities available to those who have the foresight to begin to venture down that path. Looking at the aging population trends in European countries, North America, Japan or China, we can easily conclude that one of the biggest challenges governments are facing is dealing with an increasing population over the age of 65 together with a declining birth rate. As The Independent published about grave concerns in Japan on this issue a few months ago, "Prime Minister Shinzo Abe called for a "national movement" to address Japan's demographic challenges. The government has taken steps to keep older workers in their jobs longer, and to encourage companies to invest in automation". Aging-in-place may be the key to reducing the side-effects of the phenomenon where the number of rooms in nursing homes cannot keep up with the string increase in senior population.

High-intensity storms cause billions of pounds of damage every year, and climate change is set to make this worse in future. We already appear to be seeing more frequent and intense windstorms. Hurricane Ophelia and Storm Eleanor both wreaked havoc in the British Isles over the winter, including injuries, power cuts and severe travel delays. It's not only commuters and households that are affected. Every year across Europe, the number of trees that commercial forests lose to storms is equivalent to the annual amount of timber felled in Poland.

With 400 hours of video being uploaded to YouTube every minute and fleets of self-driving cars mapping high definition 3D maps of roads all over the world, data is being created, stored and processed at a rate never seen before. Ninety percent of the world's data has been created in the last two years, according to IBM and other industry sources, and with new data hungry applications on the rise (autonomous driving, the Internet of Things, artificial intelligence), there's no sign of this trend abating. Organizing, storing and processing all that data comes with not only business, but also environmental challenges. In fact, networking and telecom equipment maker Huawei has estimated that global computing power could consume as much as 20% of global electricity in 2025 and account for 3.5% of global emissions.[1] All this data processing also requires large amounts of water to keep servers from overheating - roughly 1.8 liters for every kWh consumed - according to the U.S. Department of Energy (DOE).

Having invented the first machine gun, Richard John Gatling explained (or at least justified) his invention in a letter to a friend in 1877: With such a machine, it would be possible to replace 100 men with rifles on the battlefield, greatly reducing the number of men injured or killed. This sentiment, replacing soldiers--or at least protecting them from harm to the greatest extent possible through the inventions of science and technology--has been a thoroughly American ambition since the Civil War. And now, with developments in computing, artificial intelligence and robotics, it may soon be possible to replace soldiers entirely. Only this time America is not alone and may not even be in the lead. Many countries in the world today, including Russia and China, are believed to be developing weapons that will have the ability to operate autonomously--discover a target, make the decision to engage and then attack, without human intervention.