Sum-Product Networks (SPNs) are a class of expressive yet tractable hierarchical graphical models. LearnSPN is a structure learning algorithm for SPNs that uses hierarchical co-clustering to simultaneously identifying similar entities and similar features. The original LearnSPN algorithm assumes that all the variables are discrete and there is no missing data. We introduce a practical, simplified version of LearnSPN, MiniSPN, that runs faster and can handle missing data and heterogeneous features common in real applications. We demonstrate the performance of MiniSPN on standard benchmark datasets and on two datasets from Google's Knowledge Graph exhibiting high missingness rates and a mix of discrete and continuous features.

Despite significant progress in object categorization, in recent years, a number of important challenges remain, mainly, ability to learn from limited labeled data and ability to recognize object classes within large, potentially open, set of labels. Zero-shot learning is one way of addressing these challenges, but it has only been shown to work with limited sized class vocabularies and typically requires separation between supervised and unsupervised classes, allowing former to inform the latter but not vice versa. We propose the notion of semi-supervised vocabulary-informed learning to alleviate the above mentioned challenges and address problems of supervised, zero-shot and open set recognition using a unified framework. Specifically, we propose a maximum margin framework for semantic manifold-based recognition that incorporates distance constraints from (both supervised and unsupervised) vocabulary atoms, ensuring that labeled samples are projected closest to their correct prototypes, in the embedding space, than to others. We show that resulting model shows improvements in supervised, zero-shot, and large open set recognition, with up to 310K class vocabulary on AwA and ImageNet datasets.

Probabilistic generative models can be used for compression, denoising, inpainting, texture synthesis, semi-supervised learning, unsupervised feature learning, and other tasks. Given this wide range of applications, it is not surprising that a lot of heterogeneity exists in the way these models are formulated, trained, and evaluated. As a consequence, direct comparison between models is often difficult. This article reviews mostly known but often underappreciated properties relating to the evaluation and interpretation of generative models with a focus on image models. In particular, we show that three of the currently most commonly used criteria---average log-likelihood, Parzen window estimates, and visual fidelity of samples---are largely independent of each other when the data is high-dimensional. Good performance with respect to one criterion therefore need not imply good performance with respect to the other criteria. Our results show that extrapolation from one criterion to another is not warranted and generative models need to be evaluated directly with respect to the application(s) they were intended for. In addition, we provide examples demonstrating that Parzen window estimates should generally be avoided.

In the past, the terms business analyst and data scientist have sometimes been used interchangeably, and indeed, in a small company, the lines between the two sorts of jobs may blur. But as more and more companies look to big data for business insights, they are shifting from relying on business analysts to predict what the future of a business might look like, and moving towards using data scientists and machine learning to interpret data and predict trends. What's the difference, you might ask? While the end result of these two jobs is often similar, a business analyst and a data scientist use different tools to get there. In general, data scientists have much greater technical expertise, especially in computer programming, systems engineering, and statistics.

A former science teacher who believed in the power and possibility of online learning over two decades ago, he taught himself how to build courses in HTML on class intranets. Kevin taught one of the first hybrid, educational technology courses for teachers, for the University of Washington. And, after building countless web pages and classes on the early world wide web, he now helps develop e-learning programs, consults on virtual training'best practices' and has many interests in other internet and educational technology-related areas. Kevin finds he's now enjoying learning more from his children who are all deep into their own technology-related careers and entrepreneurial endeavors. With two new grandchildren, he's investigating more seriously the advancing new technologies in an effort to understand the knowledge and skills necessary to achieve happiness and success in a technological future.

In the Army's Emergent Leader Immersive Training Environment, or ELITE, Soldiers hone their basic counseling skills through practice with virtual humans like virtual Staff Sergeant Jessica Chen. New research aims to get robots and humans to speak the same language to improve communication in fast-moving and unpredictable situations. Scientists from the U.S. Army Research Laboratory and the University of Southern California Institute for Creative Technologies are exploring the potential of developing a flexible multimodal human-robot dialogue that includes natural language, along with text, images and video processing. "Research and technology are essential for providing the best capabilities to our Warfighters," said Dr. Laurel Allender, director of the U.S. Army Research Laboratory Human Research and Engineering Directorate. "This is especially so for the immersive and live-training environments we are developing to achieve squad overmatch and to optimize Soldier performance, both mentally and physically."

Automated deception detection is said to be one of biggest security challenges of the century. To overcome this obstacle, researchers have developed an'automated interview system' - a virtual interrogator. It can conduct interviews and determine deviations in people's physiology and behaviour via sensors. Researchers created an'automated interview systems' that conducts interviews and determines deviations in people's physiology and behavior via sensors. During the interrogation, subject's electrodermal activity (EDA) was measured, which is the most commonly used measure in a lie detection test.

And the battle to control and exploit that data is just getting started. On Monday, the Japanese carmaker Toyota announced a new subsidiary, called Toyota Connected, that will manage and mine the data collected from its vehicles, and the company said it would collaborate with Microsoft on the venture. The data collected and delivered might include mapping data, engine statistics, and records of driver behavior. Most immediately, this could mean updating vehicle features or patching bugs remotely. But the goal is also to develop new kinds of interfaces that predict a driver's intention.

Deep Learning is a shallow course that is akin to reading CliffsNotes instead of a textbook: you'll learn some terminology and be exposed to some interesting concepts but its abbreviated coverage is likely to confuse students who are new to neural networks while leaving more experienced students unsatisfied. This course seems like a rushed attempt to capitalize on the hottest buzzword in the hottest tech industry, which is a shame because it could have been a good course if it took the time to cover the topics in adequate detail. I give Deep Learning 2 out of 5 stars: Disappointing.

Research funded by the National Science Foundation may have found an ingenious solution to poaching: applying game theory and computer science to real-life situations. One of the biggest factors in why there are so many endangered animals today is poaching – a centuries-old problem. The dwindling tiger population is one of the most glaring examples of this. Whether for sport, medicine, pelts or other body parts, poaching remains a huge threat to wildlife. Patrols have long been the most direct form of human intervention in wildlife protection.