Maps of infectious disease---charting spatial variations in the force of infection, degree of endemicity, and the burden on human health---provide an essential evidence base to support planning towards global health targets. Contemporary disease mapping efforts have embraced statistical modelling approaches to properly acknowledge uncertainties in both the available measurements and their spatial interpolation. The most common such approach is that of Gaussian process regression, a mathematical framework comprised of two components: a mean function harnessing the predictive power of multiple independent variables, and a covariance function yielding spatio-temporal shrinkage against residual variation from the mean. Though many techniques have been developed to improve the flexibility and fitting of the covariance function, models for the mean function have typically been restricted to simple linear terms. For infectious diseases, known to be driven by complex interactions between environmental and socio-economic factors, improved modelling of the mean function can greatly boost predictive power. Here we present an ensemble approach based on stacked generalisation that allows for multiple, non-linear algorithmic mean functions to be jointly embedded within the Gaussian process framework. We apply this method to mapping Plasmodium falciparum prevalence data in Sub-Saharan Africa and show that the generalised ensemble approach markedly out-performs any individual method.

The tremendous success of ImageNet-trained deep features on a wide range of transfer tasks begs the question: what are the properties of the ImageNet dataset that are critical for learning good, general-purpose features? This work provides an empirical investigation of various facets of this question: Is more pre-training data always better? How does feature quality depend on the number of training examples per class? Does adding more object classes improve performance? For the same data budget, how should the data be split into classes? Is fine-grained recognition necessary for learning good features? Given the same number of training classes, is it better to have coarse classes or fine-grained classes? Which is better: more classes or more examples per class? To answer these and related questions, we pre-trained CNN features on various subsets of the ImageNet dataset and evaluated transfer performance on PASCAL detection, PASCAL action classification, and SUN scene classification tasks. Our overall findings suggest that most changes in the choice of pre-training data long thought to be critical do not significantly affect transfer performance.? Given the same number of training classes, is it better to have coarse classes or fine-grained classes? Which is better: more classes or more examples per class?

A talking monkey seems like a thing of science-fiction, cartoons, and goofy advertisements. But new research suggests monkey speech may be closer to reality than commonly thought. Macaques actually have vocal anatomy capable of human-like speech, according to a study published Friday in the journal Science Advances. "This suggests that what makes people unique among primates is our ability to control the vocal apparatus, not the apparatus itself," Thore Jon Bergman, an evolutionary biopsychologist at the University of Michigan who was not part of the research, writes in an email to The Christian Science Monitor. Researchers studying non-human primates had previously hypothesized that the shape, size, or structure of the vocal anatomy in the animals' heads might be holding them back from making the sorts of sounds that make up human speech.

IBM Watson may be able to win "Jeopardy!" The IBM Watson for Cybersecurity beta program launched this week with 40 partners around the world in an effort to help security analysts make better, faster decisions from vast amounts of data, but experts say this is the same promise offered by many other products. IBM said Watson for Cybersecurity will feature natural language processing that can help it to "understand the unique language of security." "The truth is a lot of security vendors today are attaching '[artificial intelligence]' or'cognitive' to a number of products that are really just advanced analytics or machine learning, which are also important elements that can help in the fight against cybercrime," Diana Kelley, executive security advisor for IBM Security, told SearchSecurity. "What Watson will bring to the equation that is unique is the ability to digest vast amounts of both structured data as well as all of the intelligence that exists in natural language, like blogs, white papers and research reports. For example, there are around 10,000 security research papers published each year, and 60,000 security blog posts published every month."

Deep neural networks have evolved to be the state-of-the-art technique for machine learning tasks ranging from computer vision and speech recognition to natural language processing. However, deep learning algorithms are both computationally intensive and memory intensive, making them difficult to deploy on embedded systems with limited hardware resources. To address this limitation, deep compression significantly reduces the computation and storage required by neural networks. For example, for a convolutional neural network with fully connected layers, such as Alexnet and VGGnet, it can reduce the model size by 35x-49x. Even for fully convolutional neural networks such as GoogleNet and SqueezeNet, deep compression can still reduce the model size by 10x.

So, you want to learn how to build machine learning algorithms? But where do you start? Becoming a data scientist is a really smart career move – it's possibly one of the most valuable jobs out there. That's just one of the reasons it was hailed by the Harvard Business Review as the'sexiest job of the twentieth century' back in 2012. But learning the skills you need to become a truly great data scientist, capable of building powerful machine learning systems with languages like Python and R, isn't easy.

A new artificial intelligence system can take still images and generate short videos that simulate what happens next similar to how humans can visually imagine how a scene will evolve, according to a new study. Humans intuitively understand how the world works, which makes it easier for people, as opposed to machines, to envision how a scene will play out. But objects in a still image could move and interact in a multitude of different ways, making it very hard for machines to accomplish this feat, the researchers said. But a new, so-called deep-learning system was able to trick humans 20 per cent of the time when compared to real footage. Researchers at the Massachusetts Institute of Technology (MIT) pitted two neural networks against each other, with one trying to distinguish real videos from machine-generated ones, and the other trying to create videos that were realistic enough to trick the first system. This kind of setup is known as a "generative adversarial network" (GAN), and competition between the systems results in increasingly realistic videos.

Last month, at the world's leading general management conference--the Drucker Forum in Vienna Austria--Julian Birkinshaw, Professor of Strategy and Entrepreneurship at the London Business School and Director of the Deloitte Institute of Innovation and Entrepreneurship, declared provocatively that we are living in "the Age of Agile." The full text of Julian's important talk is set out below. He also has a new book coming out next year entitled Fast/Forward: Make Your Company Fit For the Future. Earlier this week, I discussed these issues with Julian. Steve Denning: In your talk, you spoke about three possible forms of organization: bureaucracy, meritocracy and adhocracy. Could you tell us more? Julian Birkinshaw: Bureaucracy is about and occupying roles and following rules. Meritocracy is a knowledge-based view of the organization, including big data and analytics. Adhocracy is an action-based view of the organization focused on capturing opportunities, solving problems and getting results. Denning: Where does "the age of Agile" fit into this scheme?

Machine learning, a field focused on training computers to recognize patterns in data and make new predictions, is helping doctors more accurately diagnose diseases and stock analysts forecast the rise and fall of financial markets. And now materials scientists have pioneered another important application for machine learning--helping to accelerate the discovery and development of new materials. Researchers at the Center for Nanoscale Materials and the Advanced Photon Source, both U.S. Department of Energy (DOE) Office of Science User Facilities at DOE's Argonne National Laboratory, announced the use of machine learning tools to accurately predict the physical, chemical and mechanical properties of nanomaterials. In a study published in The Journal of Physical Chemistry Letters, a team of researchers led by Argonne computational scientist Subramanian Sankaranarayanan described their use of machine learning tools to create the first atomic-level model that accurately predicts the thermal properties of stanene, a two-dimensional (2-D) material made up of a one-atom-thick sheet of tin. The study reveals for the first time an approach to materials modeling that applies machine learning and is more accurate at predicting material properties compared to past models.

Although 42% of respondents to a Tech Pro Research survey said their company lacks the skills necessary to implement and support AI and machine learning, and only 28% said they had firsthand experience with the technology, 50% said their companies will adopt these within the next few years. Almost as many respondents said all the implementation and support work would be done in house when the time comes, and that their company is working to address AI and machine learning in the corporate security plan. Among companies currently using, or planning to use AI or machine learning, the top uses were research and consumer analysis. Further down the list, those currently using AI or machine learning were more interested in uses like fraud detection and market analysis, while those still in the planning stages were interested in security monitoring and office automation. For more data and analysis on corporate preparedness and uses for AI and machine learning, plus information on vendor selection, download the full report: AI and machine learning in the enterprise: Uses, organizational readiness and vendor choices.