Asia
Nissan wins converts in Japan with ¥3 million driver-assist minivan
Yusuke Goto would have crashed his Toyota Crown five years ago if the premium sedan hadn't detected the 45-year-old was veering off and righted the steering wheel. So when he found similar features in the cheaper Nissan Serena minivan, it was an easy choice. "Earlier you could only find such features in premium sedans but I have a big family and I want to make sure they are safe when we drive for a weekend getaway," Goto said on a recent visit to a Nissan showroom in Tokyo with his family. "I've become hugely interested in autonomous driving after that experience." At ¥3 million ($27,400), Nissan Motor's Serena minivan was the first model in Japan in its price bracket that offers what's known in the industry as "level 2" autonomous driving features, similar to what Tesla offers with its Autopilot function in the $80,000 Model S. A car with level 2 functionality can control steering and speed simultaneously without intervention for a short period, allowing the driver to take his hands off the wheel and foot off pedal at the same time.
SoftBank Acquires Boston Dynamics and Schaft
We knew that Masayoshi Son, founder and CEO of telecom giant SoftBank, loved robots. Now the Japanese billionaire is about to significantly expand his collection. Minutes ago, SoftBank announced that it will be acquiring Boston Dynamics and Schaft for an undisclosed sum, in order to "collaborate in advancing the development of smart robotics technologies." Boston Dynamics and Schaft were two of the nine robot companies that Google acquired in 2013 to form the core of its robotics division, headed by Android founder Andy Rubin. As far as anyone could tell, not much happened after all those companies became part of Google, and not much continued to happen through 2016, much to the frustration of roboticists everywhere.
How Google translations are getting more natural
Mumbai: Researchers are increasingly striving to help machines translate words from one language to another the way professional translators would. This implies that machines must understand the context of words and sentences, and make sense of idioms, phrases and jokes. However, despite the fact that billions of words are being translated daily by multilingual machine translation services like Google Translate, Microsoft Translator, Systran's Pure Neural Machine Translator, WordLingo, SDL FreeTranslation, China's Baidu, Russia's Yandex or Babel Fish, machines have a long way to go before they can function as fluently as humans do when speaking in, and translating, different tongues. Barak Turovsky, product lead at Google Translate--a free multilingual machine translation service from Google Inc.--understands this dilemma well. "Today, translation by machines can be likened to my five-year-old son speaking Russian.
US warplane 'shoots down' armed drone in Syria
An American F-15 warplane has shot down a pro-government drone in Syria after it fired at coalition forces, officials said, marking an escalation of tensions in the war-torn country's south. No one was hurt in Thursday's incident which occurred near the coalition's At-Tanaf garrison close to the Jordan border, Colonel Ryan Dillon, spokesman for the coalition against ISIL (Islamic State of Iraq and the Levant), told Pentagon reporters. The drone "was armed and still had weapons on it when it was fired upon by US forces from an aircraft", said Dillon. Although the weapon deployed by the drone deployed hit only dirt, the action was "clearly meant" as an attack, added Dillon. Dillon said it was not immediately clear who owned the drone.
U.S. alleges Iran linked to drone that fired on Syria forces it backs before being shot down
WASHINGTON – A drone likely connected to Iranian-supported militias fired on U.S.-backed troops near a military camp in southern Syria on Thursday, near where the U.S.-led coalition is training Syrian rebels fighting the Islamic State group, an American military spokesman said. The weapon fired by the drone did not detonate and no one was hurt, but Army Col. Ryan Dillon told reporters at the Pentagon that it was considered a direct threat and that a manned U.S. aircraft shot it down. The attack came just hours after the U.S. bombed Syrian government and allied troops inside a protected zone in that area, and marked a sharp escalation in the skirmishes between the coalition and those pro-Syrian government forces there. Dillon said this was the first time that forces supporting Damascus had attacked coalition troops in that region, which is near the training camp in Tanf, close to the border with Jordan. He declined to say who owned or operated the drone, but other officials said it was likely Iranian or Iranian-backed Hezbollah.
Recovery Guarantees for One-hidden-layer Neural Networks
Zhong, Kai, Song, Zhao, Jain, Prateek, Bartlett, Peter L., Dhillon, Inderjit S.
In this paper, we consider regression problems with one-hidden-layer neural networks (1NNs). We distill some properties of activation functions that lead to $\mathit{local~strong~convexity}$ in the neighborhood of the ground-truth parameters for the 1NN squared-loss objective. Most popular nonlinear activation functions satisfy the distilled properties, including rectified linear units (ReLUs), leaky ReLUs, squared ReLUs and sigmoids. For activation functions that are also smooth, we show $\mathit{local~linear~convergence}$ guarantees of gradient descent under a resampling rule. For homogeneous activations, we show tensor methods are able to initialize the parameters to fall into the local strong convexity region. As a result, tensor initialization followed by gradient descent is guaranteed to recover the ground truth with sample complexity $ d \cdot \log(1/\epsilon) \cdot \mathrm{poly}(k,\lambda )$ and computational complexity $n\cdot d \cdot \mathrm{poly}(k,\lambda) $ for smooth homogeneous activations with high probability, where $d$ is the dimension of the input, $k$ ($k\leq d$) is the number of hidden nodes, $\lambda$ is a conditioning property of the ground-truth parameter matrix between the input layer and the hidden layer, $\epsilon$ is the targeted precision and $n$ is the number of samples. To the best of our knowledge, this is the first work that provides recovery guarantees for 1NNs with both sample complexity and computational complexity $\mathit{linear}$ in the input dimension and $\mathit{logarithmic}$ in the precision.
iOS 11 Beta: 5 Features We're Excited To Use On The New iPhone
Apple revealed features that will be included in the upcoming iOS 11 update at WWDC this week. The iOS 11 is expected to roll out this fall, coinciding with Apple's new smartphone lineup, which will include the company's much-anticipated iPhone 8, which will be the company's 10th anniversary iPhone. Apple wants to help you out. The iOS 11 update will allow you to see floor plans, let you browse floors, view directories and search. The feature will be available at malls in select cities to start, including New York, Boston, Tokyo, Chicago and London.
Our products will be IoT and Artificial Intelligence driven: Anant Bajaj, Joint MD, Bajaj Electricals
Anant Bajaj, Joint Managing Director of Bajaj Electricals believes his organization has always been light years ahead of the competition. Way back in 2003, the company created the CIDCO Kharagar electric circle when the concept of smart lighting was but a whisper. The FMEG company, as Bajaj likes to call it has been lighting up iconic city buildings and landmarks like the CST station, Worli sea face, Rajabai clock tower and Wankhede cricket stadium. Bajaj Electricals was also the lighting partners of the Indian leg of Justin Bieber's Purpose World Tour. Six years into his role as the Joint Managing Director of BEL, Anant Bajaj is all set to drive the company into the next orbit of growth.