Welcome to the April 4, 2016 edition of ACM TechNews, providing timely information for IT professionals three times a week.
Updated versions of the ACM TechNews mobile apps are available for Android phones and tablets (click here) and for iPhones (click here) and iPads (click here).
HEADLINES AT A GLANCE
China's Companies Poised to Take Leap in Developing a Driverless Car
The New York Times (04/03/16) John Markoff; Paul Mozur; Cao Li
China could emerge as the premier country for the development and adoption of self-driving vehicles. Boston Consulting Group's Xavier Mosquet cites research suggesting China will be the world's largest market for driverless cars within 15 years. "It's not that people are more willing to use the cars in Beijing or Shanghai, it's that the economic value is much higher in China than in the U.S.," Mosquet notes. Among the Chinese companies striving to break into the autonomous car industry is Uisee Technology, a startup established by former Intel Labs China director Gansha Wu. His team, which includes experts in machine learning, computer vision, and semiconductor engineering, intends to have a tech demonstration ready for CES 2017 in Las Vegas next January. Wu says he envisions several stages toward fully driverless vehicles, including driver-assistance systems to be followed by completely driverless autos in constrained conditions, such as on private roads, fixed routes at low speed, and in controlled settings. Meanwhile, Baidu is collaborating with BMW and has secured some local governments' regulatory and infrastructure support to launch automated public transportation services in China in the next two years.
What Will it Take to Make A.I. Sound More Human?
IDG News Service (04/01/16) Katherine Noyes
Conversation fillers such as "hmm" and "uh-huh" are critical to improving communication between humans and artificial intelligence (AI), according to Carnegie Mellon University professor Alan Black, who specializes in speech synthesis and ways to make artificially intelligence speech sound more real. Although systems such as Siri and Cortana incorporate some of Black's work, the technologies simply respond to human speech. Black says the key to making AI conversations more natural are the pauses, fillers, laughs, and anticipation that help build rapport and trust. "You need mm-hmm, back channels, hesitations, and fillers, and so far our speech synthesizers can't do that," he says. Black is experimenting with using voices recorded in dialog by modeling and incorporating the variance in human responses rather than using the same response all the time. Ultimately, he says AI systems could know the user's views on certain topics so it will not say something offensive. "It's all about building this thing that's close to what humans expect and makes it easier to have this conversation," Black says.
White House Urges Federal-Local Smart Cities Collaboration
State Tech (Winter 2016) Phil Goldstein
The White House wants the U.S. government to make a greater contribution to the development of smart cities technologies, announcing a plan in September to fund $160 million in smart cities research and collaborate on implementation with the private sector. A new report from the President's Council of Advisers on Science and Technology urges the federal government to "take a more integrated approach to supporting new technologies that can improve the lives of people in cities." Describing smart cities deployments as "a...multi-trillion-dollar business opportunity," the report indicates such efforts can nurture development of new technologies, generate jobs, revive districts and parts of cities, improve residents' quality of life, and enhance urban infrastructure. The study calls on U.S. Commerce Secretary Penny Pritzker to cooperate with the heads of the departments of Housing and Urban Development, Energy, and Transportation to work with other agencies to launch the Cities Innovation Technology Investment Initiative (CITII) to encourage, manage, and support smart cities projects. The study says the CITII should mimic the model of the Transportation Department's Smart Cities Challenge. Smart Cities Council founding chairman Jesse Berst believes city governments should be prepared to participate in federally-financed competitions and team with universities and research organizations receiving federal money to research smart cities.
How Google Plans to Solve Artificial Intelligence
Technology Review (03/31/16) Tom Simonite
Google DeepMind leader Demis Hassabis envisions general artificial intelligence that learns like a human to tackle any problem as his group's overarching goal, and he says the Go-playing AlphaGo software's recent defeat of a human champion was only an initial foray. DeepMind is using a three-dimensional simulation called Labyrinth to confront its software with increasingly complicated tasks, such as navigating mazes, in order to gain knowledge and improve the software to solve more difficult challenges. The key to this process is reinforcement learning, in which software is programmed to explore new settings and adapt its behavior to increase some type of virtual incentive. Hassabis thinks reinforcement learning is the critical component in getting machine-learning software to undertake more complex tasks than are currently possible. Among the breakthroughs DeepMind needs to fulfill its objective of solving intelligence is chunking, a trick in which people plan and take actions by working with high-level concepts that conceal many details, and adapt to new situations by reassembling the "chunks," or concepts, they already know. DeepMind intends to solve this problem by studying people's brains with the help of a p
Biology Software Promises Easier Way to Program Living Cells
Nature (03/31/16) Erika Check Hayden
Massachusetts Institute of Technology (MIT) researchers have developed Cello, software that automates the design of DNA circuits for living cells. The researchers aim to help people who are not skilled biologists to quickly design working biological systems. To use the software, a user first specifies the kind of cell they are using and what they want it to do, and then types in commands to explain how these inputs and outputs should be logically connected, using a computing language called Verilog that has long been used to design silicon circuits. Cello translates this information to design the DNA sequence that, when applied to a cell, will execute the commands. The researchers, led by MIT's Christopher Voigt, currently are writing user interfaces that would return different DNA sequences for different organisms from a single program. Cello can be accessed by anyone through a Web-based interface, or by downloading the open source code. In developing Cello, the researchers created a combination of genetic components that work together as an insulator, ensuring each biological part works no matter where in the DNA sequence it is placed. The researchers tested 60 designs made using Cello, and found 45 of them worked correctly the first time.
Million-Dollar Babies
The Economist (04/02/16)
Experts in machine learning are in high demand because large technology companies use it in many activities, ranging from basic tasks such as spam-filtering and better targeting of online advertisements to futuristic projects such as self-driving cars or scanning images to identify diseases. Tech companies offer academics the opportunity to see their research pay dividends quickly, and private-sector jobs also free academics from the uncertainty of securing research grants. In addition, tech firms offer lots of computing power and large datasets, which are especially appealing to artificial intelligence (AI) researchers and are essential for modern machine learning, according to Baidu's Andrew Ng. However, the private-sector hiring spree could hurt universities that cannot offer competitive salaries. For example, Canadian universities have long been at the forefront of AI development, but that could change if their best researchers leave to work at U.S. technology companies, says University of Toronto professor Ajay Agrawal. Another potential negative impact of this trend is if AI expertise is concentrated in just a few companies, it could create an intellectual monopoly. The threat of having one company corner the market on AI influence prompted several technology executives, including Tesla's Elon Musk, to pledge more than $1 billion for OpenAI, a nonprofit initiative that will make its research public.
Robots Infiltrate Insect World to Learn Their Ways
New Scientist (04/01/16) Signe Brewster
Paris Diderot University researchers, led by researcher Jose Halloy, have developed a way to generate a robot cockroach's behavior automatically using a combination of descriptions of cockroach habits, combining models of individual movements with group activity. The researchers then used evolutionary algorithms to optimize the models. The research is based on previous work in which Halloy's team programmed robot cockroaches mostly by hand, but this method was difficult and cannot easily be adapted for use with other types of animals. The researchers tested the generated behaviors in a computer simulation in which a mixed group of 45 cockroaches and five robots had to cooperate to make a collective choice between two shelters. Generating insect-mimicking behavior automatically was a lot quicker than doing it by hand, and it led to more lifelike behavior. The researchers think the method could be used to generate behaviors for mimics of other social species, such as honeybees, fruit flies, birds, and fish. Group-level behavior matters more than individual actions, and the researchers found the movement and paths followed by the robots did not need to match those of the real cockroaches for them to fit in.
Bringing Women Back to Computer Science: UW in National Spotlight Over Efforts
Crosscut.com (03/28/16) Cambria Roth
Only about 18 percent of computer science bachelor's degrees currently are earned by women. The drop-off in participation was largely due to the narrative that computer science is for boys, a misconception that has taken a stronger hold as time has passed. The numbers remain low because of lingering gender bias, and limited support for women pursuing the male-dominated major, as well as other issues. However, the University of Washington's Department of Computer Science and Engineering (UW CSE) may have found a method that could reverse the trend. Last year, 30 percent of the bachelor degrees in computer science awarded by UW went to women, a figure that is twice the national average. The accomplishment recently garnered the first-ever award from the National Center for Women & Information Technology, which is focused on bringing more women to the field. The UW CSE program conducts outreach to K-12 teachers and students, as well as to UW students, attempting to funnel freshmen and sophomores into the introductory course, which has supportive professors that use various mechanisms to help underrepresented students succeed. "The only thing we can do is encourage highly capable people from underrepresented groups to apply to be a major and make sure they are the best," says UW CSE chair Ed Lazowska.
Hard Mathematical Problems as Basis for New Cryptographic Techniques
Ruhr-University Bochum (Germany) (04/01/16) Julia Wieler
Ruhr-University Bochum researchers have developed new cryptographic algorithms that are based on especially hard mathematical problems that could be virtually unbreakable. The algorithms, which are so efficient they can be implemented into microdevices, are based on a range of mathematical problems. One problem, known as the lattice problem, involves finding the optimal difficulty level. The researchers say to imagine a lattice to have a zero point in one specific location; the challenge is to find the point where two lattice lines intersect that is closest to the zero point. In a lattice with approximately 500 dimensions, it is impossible to solve the problem efficiently. The researchers tested various parameters that render the lattice problem simpler or harder and used that as the basis for developing a cryptographic algorithm that could be implemented in small devices. They also are developing lattice-based encryption methods, which are necessary if two parties wish to exchange a secret message. "If somebody succeeded in breaking those algorithms, he would be able to solve a mathematical problem that the greatest minds in the world have been poring over for 100 or 200 years," says Ruhr-University Bochum professor Eike Kiltz.
University Researchers Work to Develop Higher Resolution 3D Cameras
The Michigan Daily (04/01/16) Ishi Mori
University of Michigan (U-M) researchers have received a $1.2-million donation from the W.M. Keck Foundation to develop a higher-resolution three-dimensional (3D) camera. Some 3D cameras already exist in the market, but many lack sufficient resolution to create sharp 3D visuals because they use a microlens array, which results in deteriorated image quality, says U-M professor Ted Norris. The 3D camera project is based on earlier research that resulted in a graphene photodetector. "We realized that our graphene photodetector can indeed be transparent and at the same time having high sensitivity, something conventional photodetectors cannot offer," says U-M professor Zhaohui Zhong. The researchers now are trying to develop the algorithms necessary to reconstruct the images once they are taken. "The most challenging aspect of the computational imaging part of this project is the extremely large amount of data needed to represent a 3D light field and to reconstruct it from a stack of [two-dimensional] measurements collected with the transparent sensors," says U-M professor Jeffrey Fessler. The team is aiming for a single lens reflex-size camera, and want to make it so compact it could fit into a smartphone.
UCI to Launch First-of-its-Kind Official E-Sports Initiative in the Fall
UCI News (03/30/16) Laura Rico
The University of California, Irvine (UCI) is launching an official e-sports initiative this fall, which it says is the first of its kind for a public research university. UCI eSports includes a state-of-the-art arena equipped with high-end gaming PCs, a stage for League of Legends competitions, and a live webcasting studio. In addition, as many as 10 academic scholarships will be offered to the students on the team. "We hope to attract the best gamers from around the world, and our academic programs in computer gaming science, digital arts, computer science, engineering, anthropology, law, medicine, neuroscience, and behavior create a strong foundation for research and inquiry related to gaming," says UCI's Thomas Parham. He notes the gaming center will give non-team students a chance to participate in leisure activities on campus while balancing academic and research endeavors. A recent survey of UCI students found 72 percent identify as gamers and 89 percent support the creation of an e-sports team. UCI's Institute for Virtual Environments & Computer Games facilitates multidisciplinary research projects in the fields of anthropology, art, computer science, engineering, history, medicine, psychology, science, and technology. "The eSports team and the arena will ensure that UCI continues to be a leader and trendsetter in collegiate e-sports," says UCI Association of Gamers president Jesse Wang.
The 'Not Face' Is a Universal Part of Language, Study Suggests
Ohio State University (03/28/16) Pam Frost Gorder
Ohio State University researchers identified a universal facial expression interpreted across many cultures as the embodiment of negative emotion. The "not face" look is rendered as a furrowed brow, pressed lips, and raised chin, and it is identical for native speakers of English, Spanish, Mandarin Chinese, and American Sign Language (ASL). To test their hypothesis that the not face consists of three combined expressions signifying moral disagreement--anger, disgust, and contempt--the researchers filmed and photographed 158 students as they engaged in a casual dialogue with a cameraman in their native tongue. Clear grammatical markers of negation were identified in all participants, and the researchers tagged frame-by-frame images of the subjects speaking to show which muscles were moving and in which directions. Algorithms then sifted through the frames to find common patterns, revealing the not face expression no matter what language was being spoken. A computer analysis also compared the frequency of facial muscle movements to determine all subjects formed the expression at the same rate at which people speak or sign words in a sentence. In addition, ASL speakers sometimes form the not face in lieu of signing the word "not." The next step of the project is to use new algorithms to filter out and analyze facial movements without human assistance.
Abstract News © Copyright 2016 INFORMATION, INC.
To submit feedback about ACM TechNews, contact: [email protected]
Current ACM Members: Unsubscribe/Change your email subscription by logging in at myACM.
Non-Members: Unsubscribe
|
