Association for Computing Machinery
Welcome to the August 26, 2015 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).


Vint Cerf: 'Sometimes I'm Terrified' by the IoT
IDG News Service (08/25/15) Katherine Noyes

Google chief Internet evangelist and former ACM president Vint Cerf says he finds certain aspects of the Internet of Things (IoT) worrisome, particularly its integration of appliances and bug-prone software. Cerf, co-recipient in 2004 of the ACM A.M. Turing Award, acknowledges the IoT's promise to manage appliances and yield new insights into resource use, but he also cautions the IoT will make people increasingly dependent on computers, as well as programmers' ability to write quality code. "It's fraught with issues, some technical and some legislative," Cerf notes. "Who is liable when an appliance doesn't work the way it should, and what if that's a software question?" Another troubling issue Cerf raises is the potential for malefactors to gain possession of information from the IoT for nefarious uses. He says strong authentication is required to address data safety, but also vital is giving users the flexibility to share access if they want. Drawing a distinction between the right and wrong parties to share that access with is a complicating factor, and Cerf also emphasizes the need for standards. "If everything is separate and there's no standardization, we'll need different control and monitoring systems for everything," he says. "I don't want seven different hubs and systems to control all the pieces in my house."

Study Identifies New Cheating Method in MOOCs
MIT News (08/24/15) Abby Abazorius

Researchers at the Massachusetts Institute of Technology and Harvard University have detailed a new technique of cheating in massive open online courses (MOOCs), and they recommend prevention tactics. They note the method is enabled by specific elements in MOOC design, such as the ability to set up multiple accounts for free. The technique is dubbed copying answers using multiple existences online (CAMEO), which involves creating multiple accounts, including a master account that will ultimately be certified, while the others are used to "harvest" the right answers to assessment questions for the primary account. The researchers also devised an algorithm that can spot CAMEO users by seeking pairs of related accounts in which one account responds to most of the answers wrongly and the other provides mostly right answers on the first try. A review of activity time stamps can reveal when the suspected answer-harvesting account solved a specific problem shortly after the master account did. The algorithm also searches for accounts that may share IP addresses across multiple courses. To prevent MOOC cheating, the researchers suggest several strategies, including limiting assessment solutions until after assignments are due, and randomizing questions so each student receives a customized set of problems.

Researchers Tackle Issues Surrounding Security Tools for Software Developers
NCSU News (08/24/15) Matt Shipman

With funding from the U.S. National Science Foundation, North Carolina State University (NCSU) researchers are seeking to address three distinct aspects of security tools for software developers. "Our work is focused on understanding the developers who are trying to identify security vulnerabilities in their code, and how they use [or don't use] tools that can help them find those vulnerabilities," says NCSU professor Emerson Murphy-Hill. "The one thing that ties all of our work together is that we want to help give programmers the best possible tools and help them use those tools effectively." NCSU and Microsoft Research scientists have learned developers who worked on projects where security was important were not more likely than other developers to use security tools. Murphy-Hill notes people who had seen how others use the tools, and those whose bosses expected them to use them, were most likely to employ the tools. Another study focused on whether the tools supply developers with the information to ascertain if an actual problem and remediation strategy exist. The researchers found programmers were often confused by the presentation of multiple possible fixes that lacked data about the relevant pluses and minuses of each remedy. Murphy-Hill and colleagues propose the creation of evolutionary "bespoke" tools that adapt to each developer's specific skill areas.

Smooth Robot Movements Reduce Energy Consumption by Up to 40 Percent
Chalmers University of Technology (08/24/15)

An optimization algorithm developed by researchers in Sweden could potentially reduce energy consumption in robot-intensive manufacturing industries by up to 40 percent. The algorithm is designed to optimize a robot's movements in a manner that reduces acceleration and deceleration, in addition to the time a robot is at a standstill, as a robot that is at a standstill still consumes energy. "We simply let the robot move slower instead of waiting for other robots and machines to catch up before carrying out the next sequence," says Chalmers University of Technology professor Bengt Lennartson. Moreover, the optimization determines the order in which various operations are carried out without reducing total execution time. The speed and sequence of the robot is altered rather than its operation path. Safe optimization would require coordinating several robots that are moving in the same area. The optimizer will initially identify where robots may collide, as well as the entry and exit positions for each collision zone, and for each robot path. Robots can consume about half of the total energy used for production in such industries. "The goal is to make this kind of optimization standard, and included in robots from the start," notes Chalmers researcher Kristofer Bengtsson.

MIT Builds a 3D Printer That Can Use 10 Materials at Once
Computerworld (08/25/15) Lucas Mearian

Massachusetts Institute of Technology (MIT) researchers say they have built a three-dimensional (3D) printer capable of building objects with 10 photopolymer materials at once. The researchers at MIT's Computer Science and Artificial Intelligence Laboratory say they built the MultiFab 3D printer using off-the-shelf components that cost less than $7,000. By comparison, current industrial multimaterial 3D printers can handle up to three materials at once and cost as much as $250,000. The MultiFab 3D printer mixes together microscopic droplets of photopolymers that it extrudes through inkjet printheads similar to those used in office printers. It currently uses only ultraviolet-curable photopolymers hardened by a light-emitting diode lighting system, but the researchers think it could handle additional materials such as co-polymers, hydrogels, and solvent-based materials. To compensate for the relatively low quality of its off-the-shelf components, the MultiFab 3D printer uses a computer-vision system that helps the machine self-calibrate and self-correct during printing jobs. Analysts say any production-model printer based on the MultiFab design would likely cost more than the few thousand dollars touted by the researchers, but Lux Research analyst Anthony Vicari says a production model costing under $20,000 could challenge current industrial printers in the $50,000-to-$250,000 range.

A Little Light Interaction Leaves Quantum Physicists Beaming
University of Toronto (08/24/15) Sean Bettam

An experiment conducted by physicists from the University of Toronto (UT) demonstrates how interactions between light particles can be created in useful ways. The team shot a single photon at rubidium atoms cooled to a millionth of a degree above absolute zero. The photons became "entangled" with the atoms, which affected the way the rubidium interacted with a separate optical beam. The photon changed the atoms' refractive index, which caused a tiny but measurable "phase shift" in the beam. The researchers say the process represents a step toward making the essential building block of quantum computers out of pure light. The team says this could be used as an all-optical quantum logic gate, allowing for inputs, information processing, and outputs. "We've seen the effect of a single particle of light on another optical beam," says Aephraim Steinberg, a researcher in UT's Center for Quantum Information & Quantum Computing. "Normally light beams pass through each other with no effect at all. To build technologies like optical quantum computers, you want your beams to talk to one another. That's never been done before using a single photon."

To Get Girls More Interest in Computer Science, Make Classrooms Less 'Geeky'
UW Today (08/24/15) Molly McElroy

The way computer science classrooms are decorated can influence high school girls' interest in taking introductory computer science classes, suggests a new University of Washington study. Researchers studied 270 high school girls and boys, asking them to complete questionnaires about their interest in enrolling in a computer science class, their sense of belonging in such a class, and how much they thought they personally fit with computer science stereotypes. They also were shown pictures of two mock classrooms--a "stereotypical" classroom with sci-fi movie posters and computer parts, and a "non-stereotypical" classroom with potted plants and posters of abstract art and nature scenes--and asked which they preferred and if the appearance of the classroom influenced their interest in taking a computer science class. Boys showed no preference and the rooms' appearance did not affect their interest in taking a computer science class. However, 68 percent of the girls preferred the non-stereotypical classroom and girls were three times more likely to say they would be interested in taking a computer science class in the non-stereotypical classroom. "Our findings show that classroom design matters--it can transmit stereotypes to high school students about who belongs and who doesn't in computer science," says lead researcher Allison Master.

Robots Learn to Make Pancakes From WikiHow Articles
Technology Review (08/24/15) Will Knight

European researchers working on the RoboHow project have developed a robot called PR2 that is learning to make pizzas and pancakes by reading through WikiHow's written directions. The RoboHow project, which is focused on teaching robots to understand language, could make it easier for people to communicate instructions to robots, and provide a way for machines to determine how to perform unfamiliar tasks. The research also could be useful as robots become more commonplace and need to work more closely with people. "If you have a robot in a factory, you want to say, 'take the screw and put it into the nut and fasten the nut,'" says Michael Beetz, head of the Artificial Intelligence Institute at the University of Bremen, where the RoboHow project is based. After a robot learns how a specific set of instructions relates to a task, its knowledge is added to an online database called Open Ease, through which other robots can access that understanding. The robots also learn by watching videos of humans performing tasks, and studying virtual reality data when humans have performed tasks wearing gloves enabling their actions to be tracked. "Succeeding in this domain will require a tight integration of natural language, grounding the understanding via perception, and planning complex actions via manipulation algorithms," says Carnegie Mellon University professor Siddhartha Srinivasa.

How Developing and Disguising Software Bugs Can Help Cybersecurity
The Christian Science Monitor (08/24/15) Joe Uchill

The Underhanded C competition organized by Binghamton College professor Scott Craver has entered its 10th year, and its purpose is to challenge participants to disguise the most malicious software bugs in C source code. The hope is that such exercises will encourage security professionals to program code more securely and improve their software auditing ability. This year's contest differs from previous ones by being co-hosted by the Nuclear Threat Initiative, as reflected in its nuclear disarmament treaty theme, and the offering of a $1,000 jackpot in addition to bragging rights for the winner. "Our goal is to demonstrate how difficult it is to write secure software by showing off innocent-looking code that misbehaves," Craver says. The problem this year's contest focuses on concerns the vulnerability of computer-based nuclear weapon verification systems to manipulation so they can secretly transmit the location of missile sites. Craver says the best examples are malicious code flaws so well hidden they can conceivably be explained as innocent errors. He notes their prevention requires coders to practice more caution to avoid introducing security bugs into programs.

Robot Able to Quickly Adapt to Injury
EDN Network (08/21/15) Amy Norcross

Researchers at Pier and Marie Curie University (UPMC) in Paris and the University of Wyoming say they have developed an intelligent trial-and-error algorithm, which can enable robots that have suffered damage to a limb to overcome their "injury" and continue moving. UPMC's Jean-Baptiste Mouret compares the algorithm to the process of an animal testing an injured limb. "When injured, animals do not start learning from scratch. Instead, they have intuitions about different ways to behave," Mouret says. "These intuitions allow them to intelligently select a few different behaviors to try out and, after these tests, they choose one that works in spite of the injury." The algorithm works by creating a "map" of possible gaits and methods of locomotion before a robot is deployed. UPMC's Antoine Cully says the maps can be very large, noting the map for a six-legged robot includes more than 13,000 different gaits. Once the robot is injured, it begins testing the various gaits until it finds one that enables it to keep moving. The researchers say their algorithm can find a working gait in less than two minutes, and it can enable robots to recover from five different kinds of leg injuries. They note the algorithm also could enable undamaged robots to more easily traverse uneven terrain.

Supercomputers Listen to the Heart
University of Texas at Austin (08/19/15) Jorge Salazar

New supercomputer models have come closer than ever to capturing the behavior of normal human heart valves and their replacements, according to recent studies at the University of Texas at Austin (UT Austin) and Iowa State University. The studies focused on how heart valve tissue responds to realistic blood flow, and the models can help physicians make more durable repair and replacement of valves. "At the core of what we do is the development of new material models that are much more structurally and biologically informed and can actually integrate mechanisms of failure and remodeling, growth, and adaptation to altered forces that go on," says study co-author Michael Sacks at UT Austin's Institute for Computational Engineering and Sciences (ICES). ICES relies on supercomputers at the Texas Advanced Computing Center to study turbulent flow, which includes modeling the leaflet tissue-blood flow interactions through replacement human heart valves. In another study funded by the U.S. National Institutes of Health, Sacks and colleagues took what they learned about the architecture of tissue fibers of the mitral heart valve and applied it to a big-picture model that simulated function at the organ level. The goal is to develop models on how to improve the durability of surgically-repaired heart valves, which studies show can fail within as little as three to five years after repair, Sacks notes.

Augmented Reality Glasses Could Visually Encrypt Secrets
Wired (08/20/15) Andy Greenberg

University of North Carolina researchers are developing a "visual cryptography" system using augmented reality (AR) glasses to overlay one random image over another to reveal a message. The system could not be spied on by over-the-shoulder snoops. Another possible application would be overlaying a randomized number pattern on an ATM keypad to prevent spies from collecting a user's PIN by watching them enter it. The researchers first tested their system using Google Glass, but found the device's small screen too limiting and switched to AR glasses from Epson. To test the system, they asked subjects to decipher a series of overlaid images of what appeared to be collections of black and white pixels, one on a computer screen and one projected on the lenses of the AR glasses. The system worked, but its success was limited and had several limitations. Only 26 out of 30 subjects were able to successfully decrypt the messages 100 percent of the time, and they had to view the images through glasses held in a stabilization rig so they lined up. The researchers note such a system also could be compromised by a pair of AR glasses' onboard camera or Internet connection.

Argonne Pushing Boundaries of Computing in Engine Simulations
Argonne National Laboratory (08/24/15) Greg Cunningham

Argonne National Laboratory researchers are launching a new simulation project from the Virtual Engine Research Institute and Fuels Initiative (VERIFI) that will harness 60 million computer core hours to enable more effective engine simulations. The research will be conducted on MIRA, which currently is the fifth-fastest supercomputer in the world. "This has the potential to be pioneering work, because we haven't seen anyone really trying to understand these boundary conditions, model parameters, and uncertainties at this level of detail," says Sibendu Som, the project's principal investigator and principal mechanical engineer at Argonne's Center for Transportation Research. "You really need access to these types of computing resources to resolve these questions." The research will focus on investigating how multiple variables interact simultaneously to impact the functioning of an engine. MIRA will enable the Argonne researchers to simultaneously change all of the variables and simulate the impact each has on all the others. The total simulation effort will likely involve about 1 million calculations to resolve all of the variables. The study's results will be integrated into the CONVERGE engine simulation software. The research aims to produce a new generation of gasoline compression engines that operate on the basis of low-temperature combustion.

Abstract News © Copyright 2015 INFORMATION, INC.
Powered by 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