CONNECTED devices now regularly double as digital hoovers: equipped with a clutch of sensors, they suck in all kinds of information and send it to their maker for analysis. Not so the wireless earbuds developed by Bragi, a startup from Munich. They keep most of what they collect, such as the wearers’ vital signs, and crunch the data locally. “The devices are getting smarter as they are used,” says Nikolaj Hviid, its chief executive.
Bragi’s earplugs are at the forefront of a big shift in the tech industry. In recent years ever more computing has been pushed into the “cloud”, meaning networks of big data centres. But the pendulum has already started to swing: computing is moving back to the “edge” of local networks and intelligent devices.
Last year, Google’s AI researchers rolled out the Google Cloud Machine Learning Engine to make it easier for those developers with expertise in machine learning to build machine learning models for data of any size and made such APIs as Vision, Speech, Translation and Dialogflow available to be built atop pre-trained models to bring greater scale and speed to business workloads. According to Google researcher’s, the company’s Kaggle community of data scientists and machine learning experts has passed a million members and Box, Rolls Royce Marine and Kewpie are among the 10,000-plus businesses using Google Cloud AI services.
However, relatively few businesses have the skill or money to take full advantage of the capabilities in AI and machine learning, and this month Google is taking another step in addressing this gap. The company this month introduced Cloud AutoML, a plan to make machine learning services within the Google Cloud more easily accessible to both those developers with expertise in AI and engineers with fewer skills build AI systems.
We will open four local Google Hubs called “Les Ateliers Numériques” across France, run by a network of local partners from the digital sector. These physical spaces will provide a long-term Google presence in French cities … France has produced some truly heroic figures of science—like Louis Pasteur, Marie Curie, Blaise Pascal and Sophie Germain—and its educational system still produces amazing researchers. So it’s only natural that we set up a new research team in Google France around the age’s defining technology: artificial intelligence.
Archaeologists may be one step closer to decoding the mystery of the famous Dead Sea Scrolls.
Researchers from the University of Haifa in Israel have restored and deciphered one of the last untranslated Qumran Scrolls. The collection, which consists of 900 ancient Jewish manuscripts, has been shrouded in controversy since it was unearthed more than 70 years ago.
Today, we are excited to announce the hiring of three world-renowned artificial intelligence scientists, Dr. Kenneth Church, Dr. Jun Huan and Dr. Hui Xiong, and the establishment of two additional AI labs, the Business Intelligence Lab and the Robotics and Autonomous Driving Lab, as part of Baidu’s push to strengthen fundamental AI research and development.
On a not-very-sunny day in our Golden State of California, I sat down (virtually) with Dr. Hanie Sedghi to discuss the topic of ethics in Machine Learning. Born and raised in Iran, Hanie is a research scientist at Google Brain, based in Mountain View. Prior to joining Google Brain, Hanie worked at Allen Institute for Artificial Intelligence in Seattle as a research scientist for two years.
Even in science, revolutions often go far beyond reason. This year, let’s hope that scientists of all stripes — but especially social psychologists — will slow down and start approaching one another with greater respect.
For decades, the field of social psychology has captured the public imagination with high-profile research into how humans interact. Will people obey authority figures even when it involves hurting others? How do stereotypes shape human interactions? Are facial expressions of emotion universal across cultures? All of these are questions that social psychology tries to answer. But the field is in the midst of a revolution that could end up destroying new ideas before they are fully explored — a cautionary tale not just for this field, but for all of science.
Spurred by new methods and statistical techniques, a group of “revolutionaries” — scientists and Internet bloggers both inside and outside the field — have taken it upon themselves to weed out “faulty” science. In forums such as the websites Data Colada, Replicability-Index, and Statistical Modeling, Causal Inference, and Social Science, scholars are being urged to focus on replicating the results of past studies and to reconsider their own findings if subsequent research undercuts them. Done responsibly, the revolution is something all scientists could agree is fundamental to advance the field, enabling robust and verifiable discoveries about human psychology, behavior, and biology.
Mark Zuckerberg’s crusade to fix Facebook this year is beginning with a startling retreat. The social network, its chief executive said, would step back from its role in choosing the news that 2 billion users see on its site every month.
The company is too “uncomfortable” to make such decisions in a world that has become so divided, Zuckerberg explained recently.
The move was one result of a tumultuous 18-month struggle by Facebook to come to grips with its dark side, interviews with 11 current and former executives show. As outsiders criticized the social network’s harmful side effects, such as the spread of disinformation and violent imagery, vigorous internal debates played out over whether to denounce Donald Trump directly, how forthcoming to be about Russian meddling on its platform during the 2016 election, and how to fight the perception that Facebook is politically biased.
While Twitter can be used to track the diseases themselves, [Marcel] Salathe says some of the challenges mentioned by Dodge explain why the meta-analysis of vaccine acceptance makes more sense than self-reported illnesses.
“I’m not sure Twitter is the best data source for that, because people give such weird statements about themselves when they have to self diagnose,” says Salathe. “It’s not actually so much about tracking the disease itself, but rather tracking the human response to it.”
GoViral has a further advantage, explains Rumi Chunara, the NYU computer science and engineering professor who runs that project. It relies not on self-reporting, but on lab tests that definitively assess the spread of viruses and compares them to symptom reports.
“There’s a lot of opportunity, but there’s challenges as well, and I think that’s where a lot of the science could be focused,” says Chunara.
Read more: https://www.smithsonianmag.com/innovation/can-social-media-help-us-spot-vaccine-scares-predict-outbreaks-180967875/#YJFavVVHIkBz8dbo.99
Give the gift of Smithsonian magazine for only $12! http://bit.ly/1cGUiGv
Follow us: @SmithsonianMag on Twitter
Distrust of the government’s intentions toward noncitizens may be hard to overcome, research suggests, and political developments have increased levels of distrust.
Caltech and Disney Research have entered into a joint research agreement to pioneer robotic control systems and further explore artificial intelligence technologies.
The agreement creates a framework that will allow researchers and engineers at Caltech and Disney Research to easily collaborate on projects of mutual interest. The three-year agreement officially began in August 2017 with projects focused on developing robots with new autonomous movement capabilities and improving machine learning for robots on the move. The goal is to help smooth future human-machine interactions.
One advantage of intertwining an ethnographic perspective with data mining and log analysis is gaining a better understanding of people’s activities across different devices and services where cross-device and service logs are not available. More and more digital services are accessed across multiple devices, within various locales of activity, and through multiple forms of I/O with multiple interaction models. Think of experiences that move from the voice-controlled, audio-interaction service avatar in your home to your desktop and a keyboard interaction, or to your mobile and a tap-and-swipe interaction. Or think about how you move from one context to another, such as looking up the directions to a place on your phone in your living room and then moving to your car, where it’s great if the directions are loaded and ready to be audibly delivered. Taking an approach that allows us to see typical patterns of movement across locales of activity and to understand why and when those are likely to occur will help us to design truly useful human-centered predictive models.
Northeastern University, Network Science Institute
from
Boston, MA Location: International Village, Northeastern University. Ends March 12. “The Art of Networks III presents visualizations devised in the past three years covering a broad range of topics in disciplines as diverse as cosmology, genealogy, literature, music, pedagogy, and transportation networks.” [Open to the public]
Mountain View, CAGoogle is excited to announce the second annual TensorFlow Dev Summit, which will be hosted on March 30, 2018 at the Computer History Museum in Mountain View, CA. [Invites by Application]
CHTF at University of Washington, Institute for Health Metrics and Evaluation is pleased to announce a call for proposals to fund one- to two-year pilot projects to conduct novel social science research that expands the evidence base and understanding related to health and aging. Applications that leverage Global Burden of Disease (GBD) data, bring an international focus to understanding health outcomes and health disparities, or examine trends at the subnational country level are particularly welcome. Individual awards usually range from $15,000 to $75,000 each. Deadline for proposals is January 31.
For this year’s challenge, use the Spotify Million Playlist Dataset to help users create and extend their own playlists. Registration required. Timeline to be announced.
“The Cloud AI Research Challenge invites any researcher—from students to academics to employees of public and private organizations—to build AI applications on Microsoft AI services, and the two best will be awarded USD25,000.” Deadline for submissions is April 15.
How did DeepMind do it? In this essay, I’ll try to give an intuitive idea of the techniques AlphaGo Zero used, what made them work, and what the implications for future AI research are. Let’s start with the general approach that both AlphaGo and AlphaGo Zero took to playing Go.
arXiv, Computer Science > Neural and Evolutionary Computing Hojjat Salehinejad, Julianne Baarbe, Sharan Sankar, Joseph Barfett, Errol Colak, Shahrokh Valaee
from
In this paper, we present a survey on RNNs and several new advances for newcomers and professionals in the field. The fundamentals and recent advances are explained and the research challenges are introduced.
“Here we will take you through a typical data science project lifecycle using the widely-renowned Seven C’s of effective communication – Completeness, Conciseness, Consideration, Clarity, Concreteness, Courtesy and Correctness. Depending on the type of people you are interacting with, both the order and the magnitude of each of these principles would change for most impact.”
Spatial is a domain-specific language for describing hardware accelerators for use on FPGAs and other supported spatial architectures. The language is intended to be both higher level than hardware description languages (HDLs) like Verilog, VHDL, and Chisel, while also being easier to use than Altera’s OpenCL or high level synthesis (HLS) languages like Xilinx’s Vivado.
“More than a decade ago, Dan Russell, a researcher at IBM [now at Google], won fleeting attention for his email signature: ‘Join the slow email movement! Read your mail just twice each day. Recapture your life’s time and relearn to dream.’ That was quixotic even then. While some people are slow to respond to email, most of us are quick to check it.”