Recent advances in Neural Variational Inference allowed for a renaissance in latent variable models in a variety of domains involving high-dimensional data. While traditional variational methods derive an analytical approximation for the intractable distribution over the latent variables, here we construct an inference network conditioned on the symbolic representation of entities and relation types in the Knowledge Graph, to provide the variational distributions. The new framework results in a highly-scalable method. Under a Bernoulli sampling framework, we provide an alternative justification for commonly used techniques in large-scale stochastic variational inference, which drastically reduce training time at a cost of an additional approximation to the variational lower bound. We introduce two models from this highly scalable probabilistic framework, namely the Latent Information and Latent Fact models, for reasoning over knowledge graph-based representations. Our Latent Information and Latent Fact models improve upon baseline performance under certain conditions. We use the learnt embedding variance to estimate predictive uncertainty during link prediction, and discuss the quality of these learnt uncertainty estimates. Our source code and datasets are publicly available online at https://github.com/alexanderimanicowenrivers/Neural-Variational-Knowledge-Graphs.

We propose an efficient neural framework for sentence-level discourse analysis in accordance with Rhetorical Structure Theory (RST). Our framework comprises a discourse segmenter to identify the elementary discourse units (EDU) in a text, and a discourse parser that constructs a discourse tree in a top-down fashion. Both the segmenter and the parser are based on Pointer Networks and operate in linear time. Our segmenter yields an $F_1$ score of 95.4, and our parser achieves an $F_1$ score of 81.7 on the aggregated labeled (relation) metric, surpassing previous approaches by a good margin and approaching human agreement on both tasks (98.3 and 83.0 $F_1$).

Few companies had enjoyed the sort of bull run AI chipmaker Nvidia (NVDA) had been on, returning more than 1200% between June 2015 and June 2018, eventually hitting a market cap of about $175 billion by September 2018. Then everything went south – literally – as the market took a historic plunge in the fourth quarter, taking Nvidia with it. However, while many companies have bounced back, Nvidia has continued to languish, sitting at a valuation of about $88 billion, pretty much where it was circa May 2017 when we compared its AI chip technology against AMD (AMD). Now, over the last five years, the two chip manufacturers have returned almost identical value to investors, while a number of upstart startups have risen to also challenge Nvidia's supremacy with new artificial intelligence chips. In fact, it was exactly three years ago that we first introduced you to five startups building artificial intelligence chips, and then followed that up with 12 new AI chip makers in 2017.

Rafael, an Israeli defence firm, on Monday announced it had successfully demonstrated a new "automatic target recognition" capability that relies on artificial intelligence and machine learning for the newest variant of its SPICE family of guided air-to-ground bombs. The Indian Air Force is believed to have used SPICE bombs in its attack on a Jaish-e-Mohammed camp in Balakot in February. SPICE is an acronym for Smart, Precise, Impact and Cost-Effective. The SPICE munitions come in three variants: SPICE-2000, -1000 and -250, with the number denoting the weapon's weight class in pounds. The SPICE-250, which weighs around 113kg and the newest variant of the SPICE family, was the version tested with artificial intelligence technology by Rafael.

In a Nintendo Direct broadcast today, the Japanese video game giant laid out the lineup for its Switch console for the rest of 2019 and early 2020. Introduced by newly installed Nintendo of America executive Doug Bowser – no, not that Bowser – it was light on surprises, but the highlights were a new Animal Crossing game, New Horizons, and a teaser for the sequel to the phenomenally accomplished The Legend of Zelda: Breath of the Wild. Nintendo's E3 2019 presentation was dominated by giants of the Japanese video game world – not just Nintendo's own Pokémon, Smash Bros, Animal Crossing and Zelda series, but Square Enix's Dragon Quest and Mana roleplaying series and a range of anime-styled and mech-themed games. A definitive edition of Dragon Quest XI and stylish mech battler Daemon X Machina will both be out in September, and old-school Sega series Panzer Dragoon makes an unexpected return on Switch this winter. Tongue-in-cheek action game No More Heroes 3, from Tokyo's Grasshopper Manufacture, was also announced for 2020.

Collaborative robots, or cobots, have been working with humans on the factory floor for years, but when it comes to the large-scale industrial robots that can lift and move massive pieces of manufacturing, the danger to human workers is so great that the robots are bolted down to the factory floor behind fences so a human never comes near them. That is starting to change as robotics becomes more widespread across industries. Today there are, on average, 84 robots for every 10,000 workers in the U.S., according to the International Federation of Robotics. This places the U.S. second to Europe, at 99 units, and ahead of Asia, which to date averages 63 units (though the most roboticized country in the world is South Korea). While these next-generation robots are revolutionizing companies and expanding their bottom line, there is one very real caveat: Their increasing interactivity and mobility opens up the possibility of injury to human co-workers.

Robots must be smarter if they're going to pack boxes in warehouses, scan inventory in stores, and even care for the elderly. The rise of machine learning in recent years is making that possible. Steady innovation has led to robots that can independently "learn" to navigate tight corridors and grasp delicate objects without crushing them. Some of the leading American and Japanese robotics companies and investors recently gathered in Menlo Park, Calif. to discuss artificial intelligence in robotics and its impact on business. But it may require some cooperation between the U.S. and an important overseas ally.

Sumitomo Riko Co., a Nagoya-based auto parts maker, has developed a system that can determine whether a driver is holding the steering wheel, a piece of technology that could prove to be indispensable for semi-automated cars. The firm aims to start commercial production of the system -- designed to enable drivers to switch from autonomous driving to manual control safely in case of emergencies -- in the 2020 business year. The so-called Smart Rubber sensor, made of anti-vibration electrically conductive rubber material, can determine which part of the steering wheel a driver is holding by detecting a change of pressure. The auto industry is currently engaged in fierce competition to develop technology to achieve conditional automation -- Level 3 on the Society of Automotive Engineers International's scale to 5. In Level 3, cars are self-driving but a human driver must take over the wheel in emergency situations or if the system requests that the driver intervene. But self-driving mode will not be turned off unless the system determines that the driver is ready to take the wheel to avoid an accident.

FightCamp offers an interactive library of workouts available via subscription. And the Bowflex Max Trainer cardio machine incorporates artificial intelligence to help you step your home workout game up - literally. With summer on the horizon (and Instagram stories calling), I begrudgingly pulled myself out of bed early one morning in May to get some exercise before work. I stood before a stark white, freestanding boxing bag weighed down by hundreds of pounds of sand as Andre Huseman, a high-spirited personal trainer, greeted me. "Welcome back to FightCamp," the jacked fitness professional said, and within seconds he was counting down, "3…2…1."

Amazon Personalize, first announced during AWS re:Invent last November, is now available to all Amazon Web Services customers. The API enables developers to add custom machine learning models to their apps, including ones for personalized product recommendations, search results and direct marketing, even if they don't have machine learning experience. The API processes data using algorithms originally created for Amazon's own retail business, but the company says all data will be "kept completely private, owned entirely by the customer." The service is now available to AWS users in three U.S. regions, East (Ohio), East (North Virginia) and West (Oregon), two Asia Pacific regions (Tokyo and Singapore) and Ireland in the European Union, with more regions to launch soon. AWS customers who have already added Amazon Personalize to their apps include Yamaha Corporation of America, Subway, Zola and Segment.