CGTN's Mike Walter spoke with Scott Schober, president and CEO of Berkeley Varitronics Systems, about the artificial intelligence field in China.

The design is drawing attention: TechCrunch called it "the strangest smart speaker" and Quartz said the device stands out from the smart speaker crowd "for one reason." While it goes on sale in December, Baidu already has a deal to outfit the device in 100 InterContinental Beijing hotels rooms. Baidu also introduced two robots, the Raven R and Raven Q (the latter of which is still in development). The Raven R looks like a cross between Pixar's Luxo Jr. logo and a contemporary art sculpture. It's a six-jointed robot that moves based on user commands in order to improve the interactive experience and make the device seem like it's expressing emotions.

Gaussian belief propagation (BP) has been widely used for distributed inference in large-scale networks such as the smart grid, sensor networks, and social networks, where local measurements/observations are scattered over a wide geographical area. One particular case is when two neighboring agents share a common observation. For example, to estimate voltage in the direct current (DC) power flow model, the current measurement over a power line is proportional to the voltage difference between two neighboring buses. When applying the Gaussian BP algorithm to this type of problem, the convergence condition remains an open issue. In this paper, we analyze the convergence properties of Gaussian BP for this pairwise linear Gaussian model. We show analytically that the updating information matrix converges at a geometric rate to a unique positive definite matrix with arbitrary positive semidefinite initial value and further provide the necessary and sufficient convergence condition for the belief mean vector to the optimal estimate.

In this paper, we study the out-of-sample properties of robust empirical optimization and develop a theory for data-driven calibration of the robustness parameter for worst-case maximization problems with concave reward functions. Building on the intuition that robust optimization reduces the sensitivity of the expected reward to errors in the model by controlling the spread of the reward distribution, we show that the first-order benefit of little bit of robustness is a significant reduction in the variance of the out-of-sample reward while the corresponding impact on the mean is almost an order of magnitude smaller. One implication is that a substantial reduction in the variance of the out-of-sample reward (i.e. sensitivity of the expected reward to model misspecification) is possible at little cost if the robustness parameter is properly calibrated. To this end, we introduce the notion of a robust mean-variance frontier to select the robustness parameter and show that it can be approximated using resampling methods like the bootstrap. Our examples also show that open loop calibration methods (e.g. selecting a 90% confidence level regardless of the data and objective function) can lead to solutions that are very conservative out-of-sample.

Cross-entropy loss together with softmax is arguably one of the most common used supervision components in convolutional neural networks (CNNs). Despite its simplicity, popularity and excellent performance, the component does not explicitly encourage discriminative learning of features. In this paper, we propose a generalized large-margin softmax (L-Softmax) loss which explicitly encourages intra-class compactness and inter-class separability between learned features. Moreover, L-Softmax not only can adjust the desired margin but also can avoid overfitting. We also show that the L-Softmax loss can be optimized by typical stochastic gradient descent. Extensive experiments on four benchmark datasets demonstrate that the deeply-learned features with L-softmax loss become more discriminative, hence significantly boosting the performance on a variety of visual classification and verification tasks.

The Switch became the fastest-selling console in Nintendo's history just about a month after it launched, and it's showing no signs of slowing down. Nintendo's flagship console was the number one selling video game hardware device in October, according to new figures from NPD Group, a market research firm that tracks U.S. video game sales. Between the Nintendo Switch, Super Nintendo Entertainment System [SNES] Classic, and the company's line of 3DS handheld devices, Nintendo accounted for two-thirds of all video game hardware sold in the U.S. for the second month in a row. The news also comes just after Nintendo announced that it expects to sell 14 million Switch units by the end of March 2018, which is up from their initial estimate of 10 million units. The Switch's position as the top-selling game console probably won't come as much of a surprise to anyone who has tried to purchase one in recent months, particularly over the summer.

Everyone and their robot dog wants a Nintendo Switch right now. Nintendo's new hybrid console is in extremely high demand after a strong start to its life, with bonafide classics like The Legend of Zelda: Breath of the Wild and Super Mario Odyssey already out for the young system. It has not been as easy to find one as people would hope, but thanks to some Black Friday 2017 deals, you might be able to find one after you recover from turkey exhaustion. GameSpot has helpfully compiled all the deals from every major retailer that has announced its Black Friday selection so far. There is good news and bad news when it comes to securing your very own Nintendo Switch on Black Friday.

Hamas has blamed the Israeli national intelligence agency Mossad for the assassination of one of its Tunisian members after conducting an 11-month-long investigation. The Palestinian group said Mohammed al-Zawari, a commander of its armed wing the Qassam Brigades since 2006, was fatally shot outside his home multiple times while in his car near Sfax, 270km southeast of Tunis, on December 15, 2016. Hamas had set up an investigative committee in the immediate aftermath of the assassination. Speaking at a press conference in Beirut on Thursday, Mohammed Nazzal, Hamas politburo member, called the Mossad operation a "terrorist act". "Mossad is officially accused of being behind the assassination, which is not only a terrorist act, but a violation of state sovereignty," he said.

When it's not busy making billion-dollar acquisitions to expand its robotics line-up, Japanese mobile carrier SoftBank is pursuing its other hobby: smart cars. Central to this endeavour is its partner, and fellow Japan native, Honda. Last year, the two announced plans to make cars emotive using cloud-based tech based on SoftBank's Pepper robot (think Knight Rider's KITT). The fruits of that colloboration are beginning to emerge, in the guise of the auto-maker's AI-assisted NeuV and Sports EV concepts. With the clock ticking down to Honda's 2025 deadline for driverless cars, the duo are moving on to the next phase in their connected cars project, which is all about 5G.

The IT infrastructure landscape is evolving rapidly. While some may imagine infrastructure is only about stacks of hardware locked away in data centers, it comprises disruption and innovation in every area, from servers and storage to networking and software. In a recent report by McKinsey, researchers highlight trends that are giving rise to such disruption and innovation. They have identified ten trends that are already having a major impact on IT infrastructure and will bring even more disruption over the coming years. Excepert from the study given below.