Citigroup is to roll out voice recognition software to its Asian customer base, shrinking its branch network as more customers move to online and mobile banking.

ARTIFICIAL INTELLIGENCE: Can Artificial Intelligence Create the Next Wonder Material? Nicola Nosengo Nature "Instead of continuing to develop new materials the old-fashioned way -- stumbling across them by luck, then painstakingly measuring their properties in the laboratory -- Marzari and like-minded researchers are using computer modelling and machine-learning techniques to generate libraries of candidate materials by the tens of thousands." COMPUTING: Why Machine Vision Is Flawed in the Same Way as Human Vision MIT Technology Review "If machine vision and human vision work in similar ways, are they also restricted by the same limitations? Do humans and machines struggle with the same vision-related challenges? Today we get an answer thanks to the work of Saeed Reza Kheradpisheh at the University of Tehran in Iran and a few pals from around the world. These guys have tested humans and machines with the same vision challenges and discovered that they do indeed struggle with the same kind of problems."

After an up-and-down start to the year, Chinese search giant Baidu issued earnings last week that outperformed on a host of key indicators. As we've come to expect from Baidu, revenue growth remained brisk, increasing at a healthy 31% year-over-year pace to total 2.5 billion. In keeping with its recent quarters, increased spending crimped Baidu's operating profits, which grew only 2.6% compared with the first quarter of 2015. Either way, Baidu's earnings exceeded expectations on the top and bottom line. What's more, Baidu's guidance for second-quarter sales proved better than analysts anticipated, sending the company's shares up in after-hours trading the day of the announcement.

After an up-and-down start to the year, Chinese search giant Baidu (NASDAQ:BIDU) issued earnings last week that outperformed on a host of key indicators. As we've come to expect from Baidu, revenue growth remained brisk, increasing at a healthy 31% year-over-year pace to total 2.5 billion. In keeping with its recent quarters, increased spending crimped Baidu's operating profits, which grew only 2.6% compared with the first quarter of 2015. Either way, Baidu's earnings exceeded expectations on the top and bottom line. What's more, Baidu's guidance for second-quarter sales proved better than analysts anticipated, sending the company's shares up in after-hours trading the day of the announcement.

Donald Trump knows many words. He has the best words. He's going to find the best people to help him run the country. He knows many smart people. His presidency will be classy, huge, even.

Guest blog post by Xiu-Shen Wei, originally posted here. Deep Neural Networks, especially Convolutional Neural Networks (CNN), allows computational models that are composed of multiple processing layers to learn representations of data with multiple levels of abstraction. These methods have dramatically improved the state-of-the-arts in visual object recognition, object detection, text recognition and many other domains such as drug discovery and genomics. In addition, many solid papers have been published in this topic, and some high quality open source CNN software packages have been made available. There are also well-written CNN tutorials or CNN software manuals. However, it might lack a recent and comprehensive summary about the details of how to implement an excellent deep convolutional neural networks from scratch. Thus, we collected and concluded many implementation details for DCNNs. Here we will introduce these extensive implementation details, i.e., tricks or tips, for building and training your own deep networks. We assume you already know the basic knowledge of deep learning, and here we will present the implementation details (tricks or tips) in Deep Neural Networks, especially CNN for image-related tasks, mainly in eight aspects: 1) data augmentation; 2) pre-processing on images; 3) initializations of Networks; 4) some tips during training; 5) selections of activation functions; 6) diverse regularizations; 7)some insights found from figures and finally 8) methods of ensemble multiple deep networks. Additionally, the corresponding slides are available at [slide].

Steve Jurvetson has been referred to as "the smartest guy in the room," "the smartest person in Silicon Valley" and a "brainiac," among other laudatory monikers attesting to his prodigious intellect. The Internet is chock full of videos of lectures by and interviews with the venture capitalist, a partner at Draper Fisher Jurvetson. They span such topics as rockets and space, Moore's Law, machine learning, synthetic biology, technological innovation, the rich-poor gap and "the democratization of matter." That begins to reflect the breadth of Jurvetson's interests, and also his investments. Over the years, they have included companies that became transformational, from Hotmail (the Web as a platform) to Tesla (automaker as energy company).

With Congress revisiting how Pentagon units share authority under the 1986 Goldwater-Nichols Act, the intelligence agencies under the next presidential administration should likewise review their own unity of effort to become more agile and able to integrate, the top Defense intelligence official said Thursday. "The integration of intelligence of the past 15 years is a journey that is not finished," said Marcel Lettre, undersecretary of Defense for intelligence, at a banquet for agency and industry professionals in the nonprofit Intelligence and National Security Alliance. "I hope the new administration finds clear progress from the last 15 years and takes it on with a mantle of seriousness, or even sees an opportunity to redouble the effort." Lettre, who was sworn in in December to preside over a 17 billion budget, eight components and 110,000 employees, said he also hopes the next administration will "institutionalize and make irreversible" the intelligence community's digital data sharing modernization effort known as the Intelligence Community Information Technology Enterprise (pronounced "eyesight"). "Key critical data sets are the coin of the realm for the intel community," he said.

Today on In Case You Missed It: The Smart Tissue Autonomous Robot performed surgery on its own (with a human standing by) and turns out, makes such fine, consistent stitches that it actually beats those done by real counterparts. Carnegie Mellon created a wristwatch display and ring system that makes the skin of your forearm a touch pad to interact with the screen. And McDonald's made something called the McTrax placemat in the Netherland's and music folk everywhere want one, asap. We also rounded up the week's big headlines in TL;DR and hope your weekend conversations touch on whether the UAE should build an artificial mountain to get more rain. As always, please share any great tech or science videos you find by using the #ICYMI hashtag on Twitter for @mskerryd.

Convolutional Neural Networks are great: they recognize things, places and people in your personal photos, signs, people and lights in self-driving cars, crops, forests and traffic in aerial imagery, various anomalies in medical images and all kinds of other useful things. But once in a while these powerful visual recognition models can also be warped for distraction, fun and amusement. In this fun experiment we're going to do just that: We'll take a powerful, 140-million-parameter state-of-the-art Convolutional Neural Network, feed it 2 million selfies from the internet, and train it to classify good selfies from bad ones. Yeah, I'll do real work. But first, let me tag a #selfie.