Pattern analytics, deep learning, and machine learning have fueled a rapid rise in interest in GPU computing, in addition to GPU computing applications in high performance computing (HPC) and cloud-based data analytics. As a high profile example, Facebook recently contributed its "Big Sur" design to the Open Compute Project (OCP), for use specifically in training neural networks and implementing artificial intelligence (AI) at scale. Facebook's announcement of Big Sur says "Big Sur was built with the Nvidia Tesla M40 in mind but is qualified to support a wide range of PCI-e cards," pointing out how pervasive Nvidia's Tesla platform has become for AI research. Big Sur is a 4U high chassis housing a two-processor (2P) board connected to a daughter card featuring eight full-height double-width PCIe Gen3 x16 300W accelerator card slots intended to house GPU or other PCIe-based compute accelerators. The processor board and daughter card are linked via one PCIe Gen3 x16 slot in the initial implementation.

Business intelligence has gone from static reports that tell you what happened, to interactive dashboards where you can drill into information to try and understand why it happened. New big data sources, including Internet of Things (IoT) devices, are pushing businesses from those reactive analytics – whether you look back once a month to spot trends or once a day to check for problems – to proactive analytics that give you alerts and real-time dashboards. That makes better use of operational data, which is more useful while it's still current, before conditions change. "There's a demand for real-time dashboards," says Herain Oberoi from Microsoft's Cortana Analytics team. "A lot of businesses want to get the pulse of their business. But dashboards show things that have already happened."

Cancer is responsible for one-in-three deaths in Canada, according to the Canadian Cancer Society. To patients who are diagnosed, early detection can mean the difference between life and death. A microscope using AI is being touted as a powerful new instrument in the diagnostic toolkit--one that manages to snap an astounding 36 million images per second to catch cancer cells and identify their characteristics. The microscope was designed by a team at UCLA's California NanoSystems Institute, who say it's a way to identify cancer cells in patients' blood samples faster and more accurately than current methods. In a new study published in the journal Nature Scientific Reports, they describe how, using a patented microscope outfitted with a camera, they're able to photograph cells without destroying them.

Artificial Intelligence (AI) is making progress in great strides, or at least it appears so! Almost no week passes by without some major announcements of new challenges solved by AI technology or new products powered by AI. Indeed many quantifiable factors attest an unprecedented level of activity: capital investments, number of academic papers, number of products involving AI technology, they all are on a steep rise in the past five years. Computers are already very capable at some specialized tasks that require reasoning and other abilities that we typically associate with intelligence. For example, computers can play a decent game of chess or can help us order our holiday photos. Despite this genuine progress, we are still a long way from human level intelligence because our best artificial intelligence systems are not general purpose. They cannot quickly adapt to novel tasks the way most humans can do.

Scientists at Glasgow University have established a world first by cracking the communication code of our brains. Pioneering research in the field of cognitive neuroimaging has revealed how brains process what we see. The work has been led by Prof Philippe Schyns, the head of Glasgow's school of psychology, with more than a little help from Voltaire and Salvador Dali. How Dali's mind worked is a matter of continuing conjecture. But one of his works has helped unlock how our minds work.

Last fall, the Allen Institute for Artificial Intelligence in Seattle, Washington, launched a challenge to Google Scholar, PubMed, and other online search engines by unveiling a service called Semantic Scholar. The program, originally trained on 2 million papers from the field of computer science, was intended to provide a search engine, driven by artificial intelligence (AI), to actually understand--to a limited extent--the content of published literature. Its corpus has grown to 4 million papers. And today, the institute is adding a new capability to Semantic Scholar with an equally ambitious aim: measuring the influence that a scientist or organization has had on subsequent research. The tool, which focuses only on computer science for now but will expand to neuroscience by the fall and then to other subjects, can rank papers, authors, and institutions by a specific influence score. For instance, the tool finds that the most influential computer science is happening at the Massachusetts Institute of Technology in Cambridge.

Word Sense Disambiguation (WSD) aims to determine the meaning of a word in context, and successful approaches are known to benefit many applications in Natural Language Processing. Although supervised learning has been shown to provide superior WSD performance, current sense-annotated corpora do not contain a sufficient number of instances per word type to train supervised systems for all words. While unsupervised techniques have been proposed to overcome this data sparsity problem, such techniques have not outperformed supervised methods. In this paper, we propose a new approach to building semi-supervised WSD systems that combines a small amount of sense-annotated data with information from Word Sense Induction, a fully-unsupervised technique that automatically learns the different senses of a word based on how it is used. In three experiments, we show how sense induction models may be effectively combined to ultimately produce high-performance semi-supervised WSD systems that exceed the performance of state-of-the-art supervised WSD techniques trained on the same sense-annotated data. We anticipate that our results and released software will also benefit evaluation practices for sense induction systems and those working in low-resource languages by demonstrating how to quickly produce accurate WSD systems with minimal annotation effort.

ICU mortality scoring systems attempt to predict patient mortality using predictive models with various clinical predictors. Examples of such systems are APACHE, SAPS and MPM. However, most such scoring systems do not actively look for and include interaction terms, despite physicians intuitively taking such interactions into account when making a diagnosis. One barrier to including such terms in predictive models is the difficulty of using most variable selection methods in high-dimensional datasets. A genetic algorithm framework for variable selection with logistic regression models is used to search for two-way interaction terms in a clinical dataset of adult ICU patients, with separate models being built for each category of diagnosis upon admittance to the ICU. The models had good discrimination across all categories, with a weighted average AUC of 0.84 (>0.90 for several categories) and the genetic algorithm was able to find several significant interaction terms, which may be able to provide greater insight into mortality prediction for health practitioners. The GA selected models had improved performance against stepwise selection and random forest models, and provides greater flexibility in terms of variable selection by being able to optimize over any modeler-defined model performance metric instead of a specific variable importance metric.

Genetic algorithms are a well-known method for tackling the problem of variable selection. As they are non-parametric and can use a large variety of fitness functions, they are well-suited as a variable selection wrapper that can be applied to many different models. In almost all cases, the chromosome formulation used in these genetic algorithms consists of a binary vector of length n for n potential variables indicating the presence or absence of the corresponding variables. While the aforementioned chromosome formulation has exhibited good performance for relatively small n, there are potential problems when the size of n grows very large, especially when interaction terms are considered. We introduce a modification to the standard chromosome formulation that allows for better scalability and model sparsity when interaction terms are included in the predictor search space. Experimental results show that the indexed chromosome formulation demonstrates improved computational efficiency and sparsity on high-dimensional datasets with interaction terms compared to the standard chromosome formulation.

In this note, we focus on smooth nonconvex optimization problems that obey: (1) all local minimizers are also global; and (2) around any saddle point or local maximizer, the objective has a negative directional curvature. Concrete applications such as dictionary learning, generalized phase retrieval, and orthogonal tensor decomposition are known to induce such structures. We describe a second-order trust-region algorithm that provably converges to a global minimizer efficiently, without special initializations. Finally we highlight alternatives, and open problems in this direction.