Association for Computing Machinery
Welcome to the June 1, 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).


The White House Is Finally Prepping for an AI-Powered Future
Wired (05/30/16) April Glaser

The White House is adamant the government must determine how to regulate and utilize artificial intelligence (AI) technology before it gets out of control. "AI is making policy challenges already, such as how to make sure the technology remains safe, controllable, and predictable, even as it gets much more complex and smarter," says Ed Felten, deputy U.S. Chief Technology Officer. "Some of these issues will become more challenging over time as the technology progresses, so we'll need to keep upping our game." Although AI tech has yet to possess a true sense of autonomy, scholars and policy experts are worried the massive datasets such algorithms are trained on include existing biases and structural inequities. Microsoft researcher Kate Crawford emphasizes the need to instill accountability for such data, if AI is to be used in deciding such things as mortgage and job allocations, or parole for convicts. Meanwhile, University of South Carolina professor Bryant Walker Smith says with AI decision-making increasingly critical to tech still in development--such as driverless cars--regulation could hinder building safer and more responsible systems. Meanwhile, the U.S. Office of Science and Technology Policy has an inter-agency mission to determine regulation and oversight for AI already in use, such as unmanned drones.

A Mission to Bring STEM Skills, and Robots, to Children in West Africa
The New York Times (05/29/16) Dionne Searcey

University of Nebraska-Lincoln professor Sidy Ndao aims to advance science, technology, engineering, and math (STEM) skills in West Africa by organizing events such as the Pan-African Robotics Competition. The second annual contest held in Senegal last week brought together students from 25 schools to build and program robots to navigate a maze. The students used the kit-built robots to operate in models of future farms. Ndao says although West Africa has some well-equipped math schools and tech courses, the institutions sometimes focus on rote memorization instead of contextual learning, which means students seldom link theory with practical experience. Senegalese entrepreneurs and government officials are welcoming the idea of improving STEM education, based on Ndao and others' arguments that such emphasis can improve the country. Complicating this effort is a lack of electricity for schools in certain regions, and the cost of Internet access. The robot competition seeks to create a connection between Nebraska and Senegal's farming societies, and illustrate ways robots can improve their efficiency and production. "We can change our future if we learn more about technology," says Senegalese student Joanna Kengmeni.
View Full Article - May Require Free Registration | Return to Headlines | Share Facebook  LinkedIn  Twitter 

'Super Mario Brothers' Is Hard
MIT News (06/01/16) Larry Hardesty

Researchers from the Massachusetts Institute of Technology (MIT), the University of Ottawa, and Bard College found the problem of solving a level in the "Super Mario Brothers" video game is just as challenging as the most vexing problems in the PSPACE complexity class. The research follows earlier work showing the game is at least as hard as the toughest problems in NP. The previous research defined a generic video-game structure called a locked door, which must have a path through it that can be either safe to traverse or not, and a way for the player to switch the state of the path. The two possible states of the locked door means it can represent a bit of computer memory, and because it has a path through it that can be opened or closed, it can function as an element of a computational circuit. The researchers demonstrated that any computational problem could be described by locked doors threaded together in the right configuration, and if the problem is exponentially hard, then determining how to finish the level also is exponentially hard. "Super Mario Brothers" thwarted the researchers' best attempts to build a locked door, but the latest research found a way. MIT professor Erik Demaine says the work could have significant ramifications, given the mathematical similarity between video games and computational models of real-world physical systems.

Fast, Stretchy Circuits Could Yield New Wave of Wearable Electronics
UW-Madison News (05/27/16) Renee Meiller

University of Wisconsin-Madison researchers led by professor Zhenqiang Ma say they have created the fastest stretchable, wearable integrated circuits in the world, which could further the Internet of Things and a more connected, high-speed wireless world. "We've found a way to integrate high-frequency active transistors into a useful circuit that can be wireless," Ma says. "This is a platform. This opens the door to lots of new capabilities." The circuits' design was inspired by twisted-pair telephone cables, which contain two extremely small and intertwining power transmission lines in repeating S-curves. The configuration is built in two layers with segmented metal blocks, enabling the lines to stretch without inhibiting performance while also protecting them from interference. Current loss also is substantially mitigated by confinement of the electromagnetic waves flowing through the lines. The circuits currently operate at radio frequency levels up to 40 GHz, with a thickness of 25 micrometers. The researchers say their extremely small size makes them very effective in epidermal electronic systems, which could enable medical staff to monitor patients wirelessly and remotely.

JumpRoACH Is a Robotic Bug That Leaps and Flips Just Like an Insect
IEEE Spectrum (05/27/16) Evan Ackerman

Researchers at Seoul National University and the University of California, Berkeley say they have developed a new kind of jumping mechanism for robots that has the potential to scale from tiny hops to a catapult launch. The researchers added this new technology to a hexapedal crawler robot called DASH, resulting in a new running, jumping robot called JumpRoACH. The new robot weighs only 60 grams but can jump 1.6 meters in a single hop. The jumping mechanism is comprised of a diamond-shaped four-bar linkage with joints like knees. Eight latex bands are used to store energy as they stretch when the linkage is compressed, which occurs using a wire, a pulley, and a DC motor. Gears are placed between the pulley and the motor, serving as a clutch when the rotation of the motor reverses, enabling the robot to adjust its jump height. The module alone can manage nearly three meters vertically, and when it is attached to the DASH robot it can reach between 1.1 and 1.6 meters. The motor that drives the jumping mechanism also actuates an outer shell, which the robot uses to right itself after landing.

Child's Play: Australia's Newest Roboticists See Eye-to-Eye With R2-D2
QUT News (05/25/16) Debra Nowland

Queensland University of Technology's (QUT) Christina Chalmers, who specializes in the teaching and application of robotics in classrooms, says robotic coding is a growth area in a range of industries, a trend that increases the demands on educators to promote student education in robotics. "Preliminary findings from a current study have shown even pre-school students have gone beyond simply playing games with a NAO robot," Chalmers says. In 2016, coding and robotics were implemented into Queensland's state primary schools. "Research tells us that if kids don't form positive attitudes towards science, maths, and technology early in life they can find it difficult to engage later on," she says, adding robotics provides an engaging way for both students and teachers to work together. Chalmers is leading three nationwide projects to integrate robotics into classrooms. The Robotics@QUT project includes 50 Queensland schools in low socio-economic areas. Meanwhile, Chalmers is in partnership with the Association of Independent Schools of South Australia and three other Australian universities to study how humanoid robots can be integrated into the Australian curriculum. Finally, Chalmers conducted a project in conjunction with Brisbane Catholic Education and AutismCRC to develop robotic social clubs. "It provides a way of representing and understanding [science, technology, engineering, and mathematics] concepts in ways that could not be done with pen and paper," she says.

Doubling Down on Schrodinger's Cat
Yale News (05/26/16) Jim Shelton

The Schrodinger's cat paradox applies the concept of superposition in quantum physics to everyday objects, and a team of Yale University researchers has created a more exotic type of state, in which the cat simultaneously lives or dies in two boxes, to introduce entanglement to the traditional scenario. Entanglement enables a local observation to change the state of a distant object instantaneously. The Yale researchers built a device consisting of two three-dimensional microwave cavities and an additional monitoring port, all of which are connected by a superconducting, artificial atom. The "cat" is made of confined microwave light in both cavities. "This cat...doesn't stay in one box because the quantum state is shared between the two cavities and cannot be described separately," says Yale researcher Chen Wang. There also could be two small and simple Schrodinger's cats, one in each box, which are entangled. The research has potential applications in quantum computation because a quantum computer would be able to solve certain problems faster than traditional computers by exploiting superposition and entanglement. "Generating a cat in two boxes is the first step towards logical operation between two quantum bits in an error-correctible manner," says Yale professor Robert Schoelkopf.

The First Visual Search Engine for Scientific Diagrams
Technology Review (05/27/16)

Research from Po-shen Lee and colleagues at the University of Washington found graphics play an important role in the scientific process. The team downloaded 4.8 million figures from 650,000 scientific papers from the online database PubMed Central, and trained a machine-vision algorithm to recognize diagrams, photos, tables, data plots, and equations. The researchers analyzed how this distribution varied by journal, discipline, and over time, and found the most successful papers tend to have more figures. "We find a significant correlation between scientific impact and the use of visual information, where higher impact papers tend to include more diagrams, and to a lesser extent more plots and photographs," the researchers say. The research provides the foundation for an entirely new kind of science the team calls "viziometrics," the science of visual information. The researchers have made their database searchable and available online. The search engine returns diagrams, photos, and other figures related to a scientific term, enabling scientists to reason about science and data in a new way.

IT Tools Enhance Global Crop Management Strategies
Penn State News (05/25/16) Rachel Garman

Software developed at Pennsylvania State University (PSU) is being used to help sustain Uruguay's natural resources and agricultural livelihood. PSU doctoral candidate Virginia Pravia is using the computational simulation model Cycles to study soil nutrients and other data, enabling farmers and researchers to predict future crop outcomes. Working with the National Institute of Agricultural Research in Uruguay, Pravia can provide local farmers with carbon and nitrogen levels in order to develop crop management plans. "In agronomy, we have to integrate many things that are happening in the environment at the same time; for example, what's happening with the plants, the soil, and the climate," she says. "The modeling tools are very good at helping you visualize all the processes together." Using the Cycles software, farmers can enter current data, such as soil nutrient levels, and observe potential future results at the click of a button, with the software's algorithm handling the calculations. In addition to crop management, the software can be used to predict agricultural variations due to climate change. Two PSU researchers are working to combine Cycles with the Penn State Integrated Hydrologic Modeling System, which simulates nutrient transport and hydrology.

Professor Uses Computer Science to Reduce Patients' Exposure to Radiation From CT Scans
FIU News (05/25/16) Millie Acebal

Florida International University (FIU) professor Ruogu Fang is studying how to reduce radiation from computed tomography (CT) scans using math and computer technology. Fang's method involves taking CT scan images acquired at normal radiation doses and introducing "noise" in a specific mathematical way that simulates what a scan would look like at a lower radiation dose. The goal is to show high-quality images can be obtained without exposing patients to significant amounts of radiation. "My motivation is to use big data and technology to improve the quality of healthcare and reduce the risks for patients who are seeking treatment and cure," Fang says. Fang, who developed several algorithms and tested and validated them on simulated data, has shown the same information can be obtained, if not better, using her mathematical formula. Fang's research is focused on helping stroke patients, but the technology has the potential to be applied to other medical conditions, including cardiovascular diseases and kidney dysfunction. Fang says the technology also has the potential for cross-application for magnetic resonance imaging scan use.

Google AI Expert Explains the Challenge of Debugging Machine-Learning Systems
Network World (05/25/16) Steven Max Patterson

Google research director Peter Norvig compares traditional software programming to machine learning to stress the challenges of debugging and verifying that machine learning-programmed systems function as intended. Whereas traditional coding employs Boolean-based logic for this confirmation, Norvig says machine learning is a black-box method in which computers program themselves with data, generating probabilistic logic that diverges from the true-and-false tests used to verify systems programmed with traditional Boolean logic techniques. "The problem here is the methodology for scaling this [machine-learning verification] up to a whole industry is still in progress," Norvig notes, cautioning "we don't have the decades of experience that we have in developing and verifying regular software." The compelling reason for using machine learning is its productivity, but the infeasibility of traditional programming's develop-test-release verification for machine learning stems from its continuous production and acquisition of data. Norvig proposed an alternative debugging method in which machine-learning test assertions respond with assessments instead of true, false, or equal. He also acknowledges the challenge of training a machine-learning algorithm on data for which there are no set results. Norvig's solution is to use a panel of judges or a similar arrangement to determine answers without bias.

Crowdsourcing a Virtual Duke Chapel
Duke Today (05/25/16) Kara Manke

Duke University researchers are using photogrammetry, a technique for building virtual three-dimensional (3D) models based on detailed photos, to document the Duke Chapel immediately following a restoration project. "In the future, if the molding is deteriorating or we need to make some kind of repair, we will actually have a three-dimensional picture of what that piece of stone really looked like," says Duke postdoctoral researcher Edward Triplett. Instead of trying to capture the 3D world on a two-dimensional (2D) canvas, photogrammetry uses 2D images to virtually reconstruct the 3D world. A photo series that will be used for photogrammetry should have uniformity in light, exposure, and composition, so pattern-recognition software has consistent details to reference as it constructs the 3D model. "The algorithms and the graphics processors are really spiking in terms of power--we can reconstruct every nuance, every fold, every mold of the surface of the building," Triplett says. The researchers will create a 3D model of the entire building using a computer cluster recently acquired by the Research Computing center. "These new technologies are letting us capture and understand our surroundings, even when they are completely in flux," Triplett says.

Hearing Snap, Crackle, Pop May Help Heal Your Knee
Georgia Tech News Center (05/23/16) Ben Brumfield

Researchers at the Georgia Institute of Technology (Georgia Tech) are developing a knee band with microphones and vibration sensors to listen to and quantify the sounds inside the joint, which could lead to a device to help orthopedic specialists evaluate damage after an injury and monitor recovery. Georgia Tech professor Omer Inan is leading the band's development, which is sponsored by the U.S. Defense Advanced Research Projects Agency (DARPA) Biological Technologies Office. The researchers plot out the recorded audio on a graph and match it to the knee's range of motion to see precisely where in the leg's extending and bending the knee generates such sounds. Inan notes the acoustic pattern an injured knee generates is significantly less consistent than that of an uninjured knee. Inan hopes medical research will improve on his team's acoustical sensor technology so the noises produced by the knee can be decoded into useful patterns. He says the device could lead to inexpensive, wearable monitors to benefit both athletes and senior citizens, while DARPA's interest is in the technology's potential to reduce battlefield injuries. Inan says such a device could give soldiers and doctors feedback on recovering knees to help them avoid re-injury by deferring from heavy workloads when necessary.

Abstract News © Copyright 2016 INFORMATION, INC.
Powered by Information, Inc.

To submit feedback about ACM TechNews, contact:
Current ACM Members: Unsubscribe/Change your email subscription by logging in at myACM.
Non-Members: Unsubscribe