One million anonymised eye scans from Moorfields Eye Hospital will be used to train an artificial intelligence (AI) system from Google. Machine learning algorithms will scour the images for signs of diseases such as macular degeneration and diabetes-related sight loss. Moorfields is teaming up with Google's AI division DeepMind during the scheme. Previously, DeepMind faced criticism over a little-known data sharing agreement with three London hospitals. An agreement to share patient data from the Royal Free, Barnet and Chase Farm hospitals over the past five years and continuing until 2017 was revealed by the New Scientist in May.

Google DeepMind has announced its second collaboration with the NHS, working with Moorfields Eye Hospital in east London to build a machine learning system which will eventually be able to recognise sight-threatening conditions from just a digital scan of the eye. The collaboration is the second between the NHS and DeepMind, which is the artificial intelligence research arm of Google, but Deepmind's co-founder, Mustafa Suleyman, says this is the first time the company is embarking purely on medical research. An earlier, ongoing, collaboration, with the Royal Free hospital in north London, is focused on direct patient care, using a smartphone app called Streams to monitor kidney function of patients. The Moorfields collaboration is also the first time DeepMind has used machine learning in a healthcare project. At the heart of the research is the sharing of a million anonymous eye scans, which the DeepMind researchers will use to train an algorithm to better spot the early signs of eye conditions such as wet age-related macular degeneration and diabetic retinopathy.

Famous eye hospital Moorfields has agreed to give GoogleHow AI is fuelling the car industry. Read more ... »'s DeepMindHow AI is fuelling the car industry. Read more ... » access to one million anonymous eye scans as a part of a machineHow AI is fuelling the car industry. Read more ... » learningHow AI is fuelling the car industry. Read more ... » studyMicrosoft scans photos to guess what your feelings are.

The US military have launched their first experimental, fully autonomous self-driving warship, dubbed Sea Hunter, and representing a major advance in robotic warfare, which is increasingly forming the core of America's strategy to counter the Chinese and Russians, it's designed to hunt enemy submarines. The 132ft unarmed ASW Continuous Trail Unmanned Vessel (ACTUV) prototype is the naval equivalent of Google's self-driving car. Designed to cruise on the ocean's surface for two or three months at a time and with a range of over 10,000 miles it has neither a crew nor anyone controlling it remotely. And that kind of endurance and autonomy could make it a highly efficient submarine stalker at a fraction of the cost of the Navy's manned vessels. "This is an inflection point," Deputy US Defense Secretary Robert Work said in an interview, adding he hoped such ships might find a place in the western Pacific in as little as five years.

One of the most frustrating holiday experiences I ever had was when my bank decided to lock me out of my account the minute I touched down at my destination. In short, it decided my financial activity was suspicious, my attempt to withdraw cash in Italy was probably fraudulent, and took a whole load of expensive phone calls to convince it otherwise. This is precisely the kind of situation that Featurespace is helping organisations to limit. "You can't stop 100% false positives," – i.e., these untrue notifications of fraud – says the CEO Martina King, over the phone from the company headquarters in Cambridge, UK. "But we lead to a 70% reduction."

MUMBAI DELHI: A US-based research firm is predicting that India's IT services industry will lose 6.4 lakh "low-skilled" jobs to automation in the next five years, quantifying the extent of likely pain for the first time, but Indian industry experts are urging caution and point to the other side of the coin -- the creation of new jobs in large numbers. By 2021, HfS Research said that the IT industry worldwide would see a net decrease of 9% in headcount, or about 1.4 million jobs, with countries like the Philippines, the United Kingdom and the United States also taking hits. For its part, IT industry body The National Association of Software and Services Companies (Nasscom) said that the report may not have taken into account all the jobs being created by newer technologies. "Nobody's really seen what automation and robotics will really lead to. There will be some impact of automation but overall we believe that technology adoption will actually lead to more job creation across sectors," Sangeeta Gupta, senior vice president at Nasscom, told ET. "Focus on talent increasingly has to be on skill and not scale. Jobs will exist in other places also, it's not fair to say these many jobs will get eliminated."

Different versions of this problem have received is coping with uncertainties arising from limited a-considerable attention from several research communities, e.g., priori knowledge of the environment. Acquiring necessary as a "pursuit-evasion game" in game theory [13], [14], as information and achieving the overall goal are complementary a "cow-path problem" in computer science [15] or as a subtasks that require adapting the motion of a robot during "coverage problem" in control [16], [17], but its solution for mission execution, typically accompanied by minimizing a a general probability distribution or a general geometry of the performance criterion. In this work we address an Optimal region is, to a large extent, still an open question. Effective Control Problem (OCP) for a robot with fourth-order dynamics approaches for the related persistent monitoring problem based that has to find, collect and move a finite number of on estimation [18], linear programming [19] or parametric objects to a designated spot in minimum time. The objects optimization [20] have been also been proposed. OCPs with with a-priori known masses are located in a bounded twodimensional uncertainties have also been addressed by certainty equivalent space, where the robot is capable of localizing event-triggered [21], minimax [22] and sampling-based [23] itself using a state-of-the-art simultaneous localization and optimization schemes.

When sufficient labeled data are available, classical criteria based on Receiver Operating Characteristic (ROC) or Precision-Recall (PR) curves can be used to compare the performance of un-supervised anomaly detection algorithms. However , in many situations, few or no data are labeled. This calls for alternative criteria one can compute on non-labeled data. In this paper, two criteria that do not require labels are empirically shown to discriminate accurately (w.r.t. ROC or PR based criteria) between algorithms. These criteria are based on existing Excess-Mass (EM) and Mass-Volume (MV) curves, which generally cannot be well estimated in large dimension. A methodology based on feature sub-sampling and aggregating is also described and tested, extending the use of these criteria to high-dimensional datasets and solving major drawbacks inherent to standard EM and MV curves.

We consider the question of efficient estimation in the tails of Gaussian copulas. Our special focus is estimating expectations over multi-dimensional constrained sets that have a small implied measure under the Gaussian copula. We propose three estimators, all of which rely on a simple idea: identify certain \emph{dominating} point(s) of the feasible set, and appropriately shift and scale an exponential distribution for subsequent use within an importance sampling measure. As we show, the efficiency of such estimators depends crucially on the local structure of the feasible set around the dominating points. The first of our proposed estimators $\estOpt$ is the "full-information" estimator that actively exploits such local structure to achieve bounded relative error in Gaussian settings. The second and third estimators $\estExp$, $\estLap$ are "partial-information" estimators, for use when complete information about the constraint set is not available, they do not exhibit bounded relative error but are shown to achieve polynomial efficiency. We provide sharp asymptotics for all three estimators. For the NORTA setting where no ready information about the dominating points or the feasible set structure is assumed, we construct a multinomial mixture of the partial-information estimator $\estLap$ resulting in a fourth estimator $\estNt$ with polynomial efficiency, and implementable through the ecoNORTA algorithm. Numerical results on various example problems are remarkable, and consistent with theory.

Linear dimensionality reduction techniques are powerful tools for image analysis as they allow the identification of important features in a data set. In particular, nonnegative matrix factorization (NMF) has become very popular as it is able to extract sparse, localized and easily interpretable features by imposing an additive combination of nonnegative basis elements. Nonnegative matrix underapproximation (NMU) is a closely related technique that has the advantage to identify features sequentially. In this paper, we propose a variant of NMU that is particularly well suited for image analysis as it incorporates the spatial information, that is, it takes into account the fact that neighboring pixels are more likely to be contained in the same features, and favors the extraction of localized features by looking for sparse basis elements. We show that our new approach competes favorably with comparable state-of-the-art techniques on synthetic, facial and hyperspectral image data sets.