Welcome to the October 31, 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
Beyond Silicon: Squeezing More Out of Chips
The New York Times (10/30/16) John Markoff
The need to explore alternative computing methods is gaining urgency as the end of Moore's Law looms and silicon semiconductors reach their physical limits. The answer may lie with better algorithms and new hardware circuits that boost chip performance at a lower cost. "Today, you're entering this patchwork world where you are going to find a better solution for a particular problem, and that's how we're going to advance in the future," says Georgia Institute of Technology professor Thomas M. Conte. Earlier this month, researchers at Argonne National Laboratory and two universities demonstrated a programming technique for an Intel microprocessor that uses less power to execute the same work. They deactivated half of the chip's circuitry committed to mathematical precision, and then "reinvested" the savings to enhance the quality of the computed outcome. Supercomputer designers say this is a significant achievement because the high-energy needs of the fastest computers have become especially challenging as researchers attempt a transition from petaflop computers to exaflop systems. The Argonne researchers are probing the concept of trading off precision to realize dramatic computing efficiency improvements. Rice University professor Krishna V. Palem says his team plans to broaden the Argonne research to run climate change-related models more efficiently.
Boffins Predict Web Scams With Domain Registration Data
The Register (UK) (10/31/16) Richard Chirgwin
Researchers at the International Computer Science Institute, Princeton University, the University of California, Santa Barbara, Google, and the University of California, Berkeley have teamed up to create Proactive Recognition and Elimination of Domain Abuse at Time-Of-Registration (PREDATOR), an algorithm designed to analyze domain name registration behavior so it can accurately predict which names will likely be used to run online scams. The researchers estimate PREDATOR successfully flagged 70 percent of domain registrations that were later abused, while generating a false positive rate of only 0.35 percent. Princeton professor Nick Feamster says registering a series of similar names and the occurrence of registrations in bursts are both telltale signs of scammer behavior that PREDATOR is trained to recognize. Other factors the researchers say PREDATOR is keyed to examine include registration history, and a small group of registrars that scammers tend to gravitate toward. The researchers are analyzing the Domain Name Zone Alert file, and training a machine-learning model in an attempt to build a watch-list the industry can use as a first response to circumvent the relatively sluggish blacklist process. The team presented a paper on the technique last week at the ACM Conference on Computer and Communications Security (CCS 2016) conference in Vienna, Austria.
Collaboration Yields Open Source Technology for Computational Science
Oak Ridge National Laboratory (10/28/16) Scott Jones
The Eclipse Science Working Group (SWG) has released five projects to expedite scientific breakthroughs by streamlining computational science workflows. "The [SWG] has not only helped us create software to simplify complex computational science, but it has also allowed us to become a part of a thriving community that focuses on collaboration, interoperability, and cutting-edge research," says SWG chair Jay Billings. The five projects include the Eclipse Integrated Computational Environment project, a scientific workbench and workflow environment to improve the user experience for computational scientists and make it possible for developers to deploy rich interactive capabilities for their science codes and integrate many different scientific computing technologies in one common user environment. The Eclipse Advanced Visualization Project provides advanced visualization technologies for plotting, geometry, meshing, and three-dimensional visualization to address the needs of projects in the working groups. The January project is a set of libraries designed to handle scientific data in Java. The Triquetrum project delivers an open platform for managing and executing scientific workflows, and the Chemclipse project supports users who analyze data acquired from systems used in analytical chemistry. The SWG exemplifies how scientific research organizations can collaborate to deliver free software to facilitate the next major scientific breakthroughs, says Eclipse Foundation executive director Mike Milinkovich.
Robot Learns to Play With Lego by Watching Human Teachers
New Scientist (10/26/16) Aviva Rutkin
Researchers from the Freiberg University of Mining and Technology in Germany have trained a robot to play with Legos by having it observe two humans wearing motion-tracking tags as they built a Lego rocket. After just one session, the robot was able to partner with a human to build the rocket, and it could adjust when some blocks were not exactly where the robot expected to find them. The project is one of several recent examples of teaching robots through human demonstration. For example, earlier this month, Google researchers taught a robot how to open a door by physically guiding it through each step. "We have a lot of intuition about how various manipulation skills can be performed, and it only seems natural that transferring this intuition to robots can help them learn these skills a lot faster," say the Google researchers. The Freiberg team thinks learning through human demonstrations will make robots better able to assist humans with skilled factory work. "Ideally, humans and robots together should be able to do something that, individually or separately, they wouldn't have been able to do alone," says Arizona State University researcher Heni Ben Amor, who contributed to the project.
Supporting Prospective Women in STEM Starts With Accessible Mentors
The Daily (10/27/16) Aleenah Ansari
A University of Washington (UW) study examining gender disparities in science, technology, engineering, and mathematics (STEM) found an unwelcoming culture and a lack of mentors were the main deterrents to women considering STEM careers. The researchers found the differences between more and less gender-balanced fields were due to the presence of an overtly masculine culture in the classroom. They say this culture reinforced stereotypes about who is able to achieve in STEM roles, as well as a lack of relatable role models. An insufficient early exposure to courses in computer science, engineering, and physics also is a significant factor, as students who lack prior experience often are discouraged from taking introductory courses and applying to STEM programs. Mentorship activities, such as UW Engineering Discovery Days, enable current STEM students to share their projects and experiences in the field with K-12 students. UW professor Sapna Cheryan says mentorship does not have to be limited to women in the field. "Our role model studies that we have done in the past have shown that men can be just as good role models for women if they make themselves relatable," she notes. "This is something that people of all genders can participate in."
Researchers Find Weakness in Common Computer Chip
Binghamton University (10/25/16)
Haswell central-processing unit (CPU) components have weaknesses that make common computer operating systems vulnerable to malicious attacks, according to research by a team from Binghamton University and the University of California, Riverside. The researchers say computer hackers can manipulate a CPU's branch predictor, exploit a weak point in address space layout randomization (ASLR) software, and take control of individual, company, and government computers. The team demonstrated the weakness in commonly-used Linux operating systems using Intel processors, but they say the vulnerability could extend to other operating systems, such as Windows and Android. "Previous research demonstrated several ways to bypass ASLR, but our attack is just more efficient and direct," says Binghamton professor Dmitry Ponomarev. "It does not change the fundamental state of the security arms race. Individual users should not worry about this attack, but rather make sure that operating systems are always updated to ensure that other exploitable vulnerabilities are not present." The team identified several ways to mitigate the attacks in a paper presented earlier this month at the IEEE/ACM International Symposium on Microarchitecture (MICRO-49) in Taipei, Taiwan.
Finding Patterns in Corrupted Data
MIT News (10/26/16) Larry Hardesty
A new set of algorithms developed by a multi-university team of researchers is capable of efficiently model-fitting probability distributions to high-dimensional data. "From the vantage point of theoretical computer science, it's much more apparent how rare it is for a problem to be efficiently solvable," says Massachusetts Institute of Technology professor Ankur Moitra. "If you start off with some hypothetical thing--'Man, I wish I could do this. If I could, it would be robust'--you're going to have a bad time, because it will be inefficient. You should start off with the things that you know that you can efficiently do, and figure out how to piece them together to get robustness." Moitra and his collaborators developed an algorithm whose running time rises with the number of data dimensions at a more reasonable rate than algorithms that take two-dimensional cross-sections of the data graph to see if they resemble Gaussian distributions. Their algorithm hinges on what metric to use when quantifying how far off a dataset is from a range of distributions with approximately the same shape. Another key to its performance is identifying the regions of data in which to start taking cross sections.
A Complete Waste of Energy
UNews (UT) (10/25/16) Vincent Horiuchi
A new type of switch for electronic circuits could enable smartphones and laptops to run at least twice as long on a single battery charge, and newer all-digital appliances to be much more power efficient. Developed by University of Utah engineers led by professor Massood Tabib-Azar, the microscopic electronic switches can grow and dissolve wires inside the circuitry of appliances and devices that instantly connect and disconnect electrical flow. Tabib-Azar says the switch uses solid electrolytes such as copper sulfide to grow a wire between two electrodes when an electrical current passes through them, turning the switch on. Reversing the polarity of the electrical current turns it off. A third electrode is used to control the process of growing and breaking down the wire. The researchers note another advantage of the technology is it would produce less heat in the appliance or device, helping ensure the reliability of their components over time. The time it takes to grow and break down the wires makes the process slower than typical switches in regular silicon-based electronics, but Tabib-Azar expects this to improve as the researchers continue to optimize the process.
New UTSA Study Describes Method to Detect Dishonesty Online
UTSA Today (10/24/16) Joanna Carver
A method for detecting people dishonestly posting online content across multiple accounts is described in a study conducted by professor Kim-Kwang Raymond Choo at the University of Texas at San Antonio. The statistical technique analyzed multiple writing samples from the most prolific online commenters on various news websites. Choo says the algorithm determined many people expressing their views online were actually all connected to a few singular writers with multiple accounts. This practice, known as "astroturfing," is legal but ethically dubious, according to Choo. Astroturfing is employed by businesses to manipulate social media users or online shoppers, by having one paid associate post bogus reviews on sites about products for sale. On social media, astroturfers set up several false accounts to espouse opinions, creating an illusory consensus when in fact one person is masquerading as many. Choo says he is currently investigating whether the new algorithm can be used to curb plagiarism and reduce cheating. "In addition to raising public awareness of the problem, we hope to develop tools to detect astroturfers so that social media users can make informed choices and resist online social manipulation and propaganda," he says.
New Professor Creates Self-Folding, Origami Robots
Northeastern University News (10/24/16) Molly Callahan
Northeastern University professor Sam Felton has developed printable, foldable robots with the goal of getting them to fold themselves. "The idea is to push the boundaries of what's possible in self-folding structures," Felton says. Felton has worked with other researchers to further explore the method he established for building self-folding machines, an idea based on origami. Felton's robots are made of paper that is sandwiched between layers of pre-stretched polystyrene. The self-folding robots use a heat-contracting mechanism in which strips of copper are placed along the fold lines of the robot. Microchips installed on the robot heat up the copper strips by sending electric current through them. When the copper is heated up, the polystyrene material contracts, causing the joint to buckle and fold. The researchers want to find a way to "scale up" the robots, which involves moving away from using heat as the catalyst for folding. "I'm really interested in expanding it to the very large scale, where it could be useful both for architecture and buildings that could assemble themselves as well as for space exploration, where it's very difficult to transport stuff up into orbit," Felton says.
Researchers Are Teaching Artificial Intelligence How to Terrify Humans
NextGov.com (10/25/16) Adrienne LaFrance
Massachusetts Institute of Technology (MIT) researchers have trained a computer to produce scary images as part of a deep-learning project called the Nightmare Machine. "Creating a visceral emotion such [as] fear remains one of the cornerstones of human creativity," the researchers say. "This challenge is especially important in a time where we wonder what the limits of artificial intelligence are: can machines learn to scare us?" The team fed a body of terrifying images to an algorithm to evoke certain distinct aesthetics. When presented with ordinary pictures, the algorithm would use its acquired knowledge to tweak the images in spooky ways, such as transforming a photo of the Statue of Liberty under a clear sky into an image of the landmark in the midst of a ghostly storm. The Nightmare Machine is even more adept at generating ghoulish images of disfigured human faces. The MIT researchers are asking people to vote on which of the algorithm-produced faces is most terrifying so they can refine it.
You Are Less Anonymous on the Web Than You Think--Much Less
Stanford University (10/20/16) Vignesh Ramachandran
Abstract News © Copyright 2016 INFORMATION, INC.
To submit feedback about ACM TechNews, contact: [email protected]