We evaluate the uncertainty quality in neural networks using anomaly detection. We extract uncertainty measures (e.g. entropy) from the predictions of candidate models, use those measures as features for an anomaly detector, and gauge how well the detector differentiates known from unknown classes. We assign higher uncertainty quality to candidate models that lead to better detectors. We also propose a novel method for sampling a variational approximation of a Bayesian neural network, called One-Sample Bayesian Approximation (OSBA). We experiment on two datasets, MNIST and CIFAR10. We compare the following candidate neural network models: Maximum Likelihood, Bayesian Dropout, OSBA, and --- for MNIST --- the standard variational approximation. We show that Bayesian Dropout and OSBA provide better uncertainty information than Maximum Likelihood, and are essentially equivalent to the standard variational approximation, but much faster.

The three operator splitting scheme was recently proposed by [Davis and Yin, 2015] as a method to optimize composite objective functions with one convex smooth term and two convex (possibly non-smooth) terms for which we have access to their proximity operator. In this short note we provide an alternative proof for the sublinear rate of convergence of this method.

Text documents can be described by a number of abstract concepts such as semantic category, writing style, or sentiment. Machine learning (ML) models have been trained to automatically map documents to these abstract concepts, allowing to annotate very large text collections, more than could be processed by a human in a lifetime. Besides predicting the text's category very accurately, it is also highly desirable to understand how and why the categorization process takes place. In this paper, we demonstrate that such understanding can be achieved by tracing the classification decision back to individual words using layer-wise relevance propagation (LRP), a recently developed technique for explaining predictions of complex non-linear classifiers. We train two word-based ML models, a convolutional neural network (CNN) and a bag-of-words SVM classifier, on a topic categorization task and adapt the LRP method to decompose the predictions of these models onto words. Resulting scores indicate how much individual words contribute to the overall classification decision. This enables one to distill relevant information from text documents without an explicit semantic information extraction step. We further use the word-wise relevance scores for generating novel vector-based document representations which capture semantic information. Based on these document vectors, we introduce a measure of model explanatory power and show that, although the SVM and CNN models perform similarly in terms of classification accuracy, the latter exhibits a higher level of explainability which makes it more comprehensible for humans and potentially more useful for other applications.

Over time, we expect the distinction between'AI' companies and other software providers to blur and then disappear, as machine learning is employed to tackle a wide variety of business processes and sectors. Today, however, it is possible to point to a sub-set of early stage software companies defined by their focus on AI. We've researched 226 early stage AI companies in the UK and met with 40 of them. We've developed a map (Figure 1, above) to place the 226 according to their: Purpose: Is the company focused on improving a business function (for example, marketing or human resources) or a sector (healthcare, education, agriculture)? Or does the company develop an AI technology with cross-domain application?

A new report from the White House warns that millions of jobs could be automated out of existence in coming years. But it cautions against one much discussed solution: giving away free money. The report, published this week by the President's Council of Economic Advisers, joins a growing body of work forecasting massive jobs losses due to automation and artificial intelligence. A paper published in 2013 by Oxford University researchers, for example, estimated that as many as 47 percent of all jobs could eventually be automated. The new report, likewise, forecasts millions of job losses in careers such as truck driving, as self-driving vehicles hit the roads, as well as low-skilled jobs.

Thirty of the world's top scientists are scheduled to meet at UC San Diego in February to discuss the toughest challenges in robotics and automation, including how to make driverless cars safe for a mass audience. The experts are being brought together by Henrik Christensen, the prominent Georgia Tech engineer who was hired in July to run UC San Diego's young Contextual Robotics Institute. Christensen said at the time, "I want to build a research institute that, ideally, will be in the top five in the world five years from now. Why not see if we can make San Diego'Robot Valley.'" The February forum is being eyed as a step toward raising the university's visibility in robotics, a field defined by grand advances and embarrassing setbacks.

It's no longer surprising to see postal services experimenting with delivering mail using drones. However, France is kicking things up a notch: its national mail service will be the first to deliver packages by drone on a regular route. DPDgroup, the express courier subsidiary of the mail service, is running a test program where a hexacopter drone (not shown here) will carry packages up to 6.6 pounds along a 9.3-mile route in France's southern Provence region. These are commercial customers using dedicated spaces to collect their orders, so you can't strictly compare this to Amazon's recent home delivery in the UK. It would be considerably more challenging to deliver to individuals, who don't have the luxury of secure areas or readily available staff.

ASUS put a choir of 10 Zenbo robots together for gadget's official launch The bots sang and danced to Mariah Carey's'All I Want For Christmas Is You' Pre-orders start Jan. 1 and units will cost between $620 and $780 ASUS put a choir of 10 Zenbo robots together for gadget's official launch The bots sang and danced to Mariah Carey's'All I Want For Christmas Is You' Pre-orders start Jan. 1 and units will cost between $620 and $780 A Chinese firm launched its new robot in a way that is sure to spread some holiday cheer. Ten Zenbos took the stage in Taiwan and wowed the audience by signing Mariah Carey's'All I Want For Christmas Is You' in a festive, and somewhat terrifying, spectacle The incredible eight storey vending machine that dispenses... Is your child lying to you about still believing in Santa?... Apple's iPhone 8 may have wireless charging that works up to... Check your Groupon account now! Hacks could mean millions of... The incredible eight storey vending machine that dispenses... Is your child lying to you about still believing in Santa?... Apple's iPhone 8 may have wireless charging that works up to... Check your Groupon account now! Hacks could mean millions of...

Otsimo provides free games for autistic children to help improve attention spans using machine learning and gamification. Autistic spectrum disorders, or ASDs, impose huge costs, both human and economic, on sufferers, their families, and the community. The human toll, in terms of care and impaired relationships, is almost impossible to quantify. Looking at the US alone, the economic expenditure last year was estimated at about $268bn. The exact causes of the pathology are still unknown, and there is no medical solution except for early and intensive education, which can at least reduce the symptoms.

The Jeopardy Challenge helped us address requirements that led to the design of the DeepQA architecture and the implementation of Watson. After 3 years of intense research and development by a core team of about 20 researcherss, Watson is performing at human expert levels in terms of precision, confidence, and speed at the Jeopardy quiz show. Our results strongly suggest that DeepQA is an effective and extensible architecture that may be used as a foundation for combining, deploying, evaluating, and advancing a wide range of algorithmic techniques to rapidly advance the field of QA. The architecture and methodology developed as part of this project has highlighted the need to take a systems-level approach to research in QA, and we believe this applies to research in the broader field of AI. We have developed many different algorithms for addressing different kinds of problems in QA and plan to publish many of them in more detail in the future.