It was December 2012, and Doug Burger was standing in front of Steve Ballmer, trying to predict the future. Ballmer, the big, bald, boisterous CEO of Microsoft, sat in the lecture room on the ground floor of Building 99, home base for the company's blue-sky R&D lab just outside Seattle. The tables curved around the outside of the room in a U-shape, and Ballmer was surrounded by his top lieutenants, his laptop open. Burger, a computer chip researcher who had joined the company four years earlier, was pitching a new idea to the execs. He called it Project Catapult. The tech world, Burger explained, was moving into a new orbit.

If I had to pick one area I find most exciting, most potentially disruptive and thus the best opportunity for return on investment, it would be A.I. Or is it the other way around? The anatomy of infrastructure is not unlike that of the human body, and we all know it's possible to live long, healthy lives. But we also know that systems out of balance in the body lead to sickness and disease (or, in the case of Industrial Revolution, economic disease and monetary sickness). It hasn't happened yet, as history has shown that human behavior, fear, and greed are more sure in life than death and taxes. Remember, we're trying to construct an infrastructure with systems working in harmony: the Anatomy of the (Healthy) Future Clearly, the best representation of the nervous system in our cyber-anatomy infrastructure is Artificial Intelligence, the future of which is Deep Learning (Augmented Intelligence, Machine Learning, and Neural Networks have actually been around some time).

The literature on judgment aggregation is moving from studying impossibility results regarding aggregation rules towards studying specific judgment aggregation rules. Here we give a structured list of most rules that have been proposed and studied recently in the literature, together with various properties of such rules. We first focus on the majority-preservation property, which generalizes Condorcet-consistency, and identify which of the rules satisfy it. We study the inclusion relationships that hold between the rules. Finally, we consider two forms of unanimity, monotonicity, homogeneity, and reinforcement, and we identify which of the rules satisfy these properties.

Many machine learning tasks can be formulated as Regularized Empirical Risk Minimization (R-ERM), and solved by optimization algorithms such as gradient descent (GD), stochastic gradient descent (SGD), and stochastic variance reduction (SVRG). Conventional analysis on these optimization algorithms focuses on their convergence rates during the training process, however, people in the machine learning community may care more about the generalization performance of the learned model on unseen test data. In this paper, we investigate on this issue, by using stability as a tool. In particular, we decompose the generalization error for R-ERM, and derive its upper bound for both convex and non-convex cases. In convex cases, we prove that the generalization error can be bounded by the convergence rate of the optimization algorithm and the stability of the R-ERM process, both in expectation (in the order of $\mathcal{O}((1/n)+\mathbb{E}\rho(T))$, where $\rho(T)$ is the convergence error and $T$ is the number of iterations) and in high probability (in the order of $\mathcal{O}\left(\frac{\log{1/\delta}}{\sqrt{n}}+\rho(T)\right)$ with probability $1-\delta$). For non-convex cases, we can also obtain a similar expected generalization error bound. Our theorems indicate that 1) along with the training process, the generalization error will decrease for all the optimization algorithms under our investigation; 2) Comparatively speaking, SVRG has better generalization ability than GD and SGD. We have conducted experiments on both convex and non-convex problems, and the experimental results verify our theoretical findings.

Tuplejump, which is based in India, was acquired by Apple in June, according to a report.

They are the APAC and Australian leaders of some of the largest, or fastest rising marketing tech, adtech and ecommerce companies. And they are passionate about helping their clients understand their own consumers using data analytics. So we figured it was time to turn the lens around. We used IBM's Personality Insight services in the Watson Developer Cloud to tell us a little bit about the personality of each of the following executives; Karen Stocks from Twitter, Ben Sharp from AdRoll, Liam Walsh from Amobee, Jodie Sangster from ADMA, Paul Robson from Adobe, Derek Laney from Salesforce, Paul Cross from Oracle, Matt Barrie from Freelancer and Ruslan Kogan from Kogan. Given their commitment to the cause of data-driven marketing we are sure they won't mind at bit.

The Pentagon's oft-repeated line on artificial intelligence is this: we need much more of it, and quickly, in order to help humans and machines work better alongside one another. But a survey of existing weapons finds that the U.S. military more commonly uses AI not to help but to replace human operators, and, increasingly, human decision making. The report from the Elon Musk-funded Future of Life Institute does not forecast Terminators capable of high-level reasoning. At their smartest, our most advanced artificially intelligent weapons are still operating at the level of insects … armed with very real and dangerous stingers. So where does AI exist most commonly on military weapons?

Albert Einstein spent the fall of 1915 overeating and sleeping in. He'd discovered a flaw in his theory of gravity, and competitors circled overhead. So Einstein relaxed a bit, went back to the blackboard and, months later, wrote the equation that rules the universe. More than a century later, today's Einsteins - the developers who've changed our everyday life with apps - also find themselves at an impasse. With the consumer experience evolving through artificial intelligence (AI) in ways big and small - from tastemaking playlists to facial recognition - the enterprise now faces a competitive imperative to integrate AI into the apps their products and services run on.

All those cars on California's famously gridlocked highways could be doing more than using energy. They could be producing it. The California Energy Commission is investing 2 million to study whether piezoelectric crystals can be used to produce electricity from the mechanical energy created by vehicles driving on roads. The commission is choosing a company or university to take on small-scale field tests. It will study how the small crystals, which generate energy when compressed, could produce electricity for the grid if installed under asphalt.

They were dismissed as stupid and primitive by the archaeologists who first studied their fossilised bones, but it seems Neanderthals may have been just as chatty as our own species. Researchers have discovered evidence that suggests Neanderthals may have used vocal communication just like our own species Homo sapiens. They used 3D scans to analyse the delicate ear bones from both the fossilised remains of Neanderthals and modern humans. Scientists have found evidence that suggests Neanderthals (reconstruction pictured) may have communicated vocally in similar ways to modern humans. Using highly detailed'micro-CT' scans they were able to reconstruct how the ear bones, or ossicles, would have functioned when Neanderthals were alive.