Association for Computing Machinery
Welcome to the June 8, 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).


Korean Robot Makers Walk Off With $2 Million Prize
The New York Times (06/06/15) John Markoff

This weekend, a team from the Korea Advanced Institute of Science and Technology won first place in the finals of the U.S. Defense Advanced Research Projects Agency's (DAPRA) Robotics Challenge, taking home the $2-million prize. The second- and third-place winners, the Florida-based Institute of Human and Machine Cognition and the National Robotics Engineering Center's Tartan Rescue, received $1 million and $500,000, respectively. The winning team employed one of the latest of the Hubo family of humanoid robots designed by roboticist JunHo Oh. The robot was able to kneel and drive on wheels mounted on the sides of its legs. The finals of the Robotics Challenge involved eight tasks, including driving a vehicle, opening a door, operating a portable drill, and climbing stairs. The finals added several complications, such as requiring that the robots be controlled remotely and utilize only battery power. DARPA also occasionally degraded wireless signals to more accurately simulate a crisis situation. The contest drew teams China, Hong Kong, Germany, Italy, and Japan, as well as from the U.S. and South Korea. The competition, held at the Los Angeles County Fairgrounds, also was accompanied by an exhibition that drew dozens of companies, national laboratories, and universities that demonstrated a variety of robots for attendees.
View Full Article - May Require Free Registration | Return to Headlines | Share Facebook  LinkedIn  Twitter 

Modeling Data
Penn State News (06/08/15) Moriah Nastasi

Pennsylvania State University professor Heng Xu is collecting and analyzing data in order to understand the needs, motivations, and behaviors of the fashion industry, retailers, and consumers. Xu's research focuses on the intersection of fashion and information systems. Xu says her goal is to help the average person stay in touch with the fashion world. Xu and her colleagues analyzed 15 years of data from fashion websites, magazines, runway reviews, and major fashion designers' social media accounts to find patterns. The researchers tracked groups of users who retweeted certain tweets to determine whether or not patterns exist. "We can use social media data to identify consumer engagement patterns and predict regional demand," Xu says. After the researchers complete their analyses, they plan to launch a website to publish their discoveries. "The site will include an in-depth breakdown of each season's trends, such as specific designers and fabrics that are most popular during that time," Xu says. In addition, she says the website will be user friendly so anyone can search for and understand what the latest fashion trends are.

Daphne Koller on the Future of Online Education
The Wall Street Journal (06/05/15) Alexandra Wolfe

Coursera co-founder Daphne Koller, former professor of artificial intelligence at Stanford University, hopes her online education platform will expand globally as more universities see online education evolving into a necessary, and inevitable, complement to traditional learning. Although Koller, who in 2008 received the first-ever ACM-Infosys Foundation Award in the Computing Sciences, does not envision the obsolescence of traditional university education, she says online education "gives people the opportunity to pause and reflect and grapple with the material," as well as offering valuable feedback to instructors. Universities participating in Coursera--119 in total so far--share approximately 1,000 courses free of charge on the platform, and later this year the site will open admissions for its first online MBA program from the University of Illinois. Each course available on Coursera is distinctive, demanding a blend of video lectures, assignments, and tests. The average course takes four to six weeks to complete, and course certificates that graduates can add to their resumes cost $50 to $95. Universities earn half of all revenues generated by their own courses, while Coursera's keystroke-identifying methods can help prevent users from cheating. Koller believes a new educational model will emerge in which information is supplied online and classrooms are used for interactive experiential learning instead of for lecturing, with the ultimate effect of better learning outcomes.
View Full Article - May Require Paid Subscription | Return to Headlines | Share Facebook  LinkedIn  Twitter 

Helping Robots Handle Uncertainty
MIT News (06/02/15) Larry Hardesty

Massachusetts Institute of Technology (MIT) researchers have demonstrated the use of decentralized partially observable Markov decision processes (Dec-POMDPs) to model autonomous robots' behavior in scenarios in which their communication with each other and their assessments about their surroundings are uncertain. The researchers found Dec-POMDPs are capable of determining the best way to patch together existing, lower-level robotic control systems to perform collective tasks. Because even accurate sensor readings by a robot are no guarantee against error, the robot must build a probability distribution of possible locations on top of the probability distribution of readings; this can be done by decomposing the DecPOMDP into two problems involving graphs. The first graph features nodes representing a belief state, or a probabilistic estimate of an agent's own state and the state of the world; a set of control procedures to move the agent between belief states are then generated. A second graph is then assembled for each agent, with modes representing macro-actions defined in the previous step, and the edges being transitions between macro-actions, given observations. Another MIT research team has developed a system that can produce the lower-level control systems from the bottom-up while also addressing Dec-POMDP models in a reasonable interval.

America's Unwanted Ivy Leaguers Are Flocking to India
Bloomberg (06/02/15) Dina Bass

India's best and brightest students increasingly are being drawn back to their homeland, or not leaving in the first place, as the country's technology boom takes off. This development reverses a decades-long trend of Indians coming to study at top American universities with the hope of ultimately finding employment in the U.S. University of North Carolina, Pembroke professor Sonali Jain says the Indian tech boom, as well as the constraints of the U.S. H-1B visa program that many Indians would use to remain in the U.S., are encouraging more homecomings. The Indian tech sector, especially in the e-commerce and mobile spaces, has been rapidly gathering momentum. Venture capital investments in Indian tech companies reached $1.9 billion in the fourth quarter, nearly six times what they were a year before. Meanwhile, financial institutions, including hedge funds, investment firms, and asset managers, have poured $3.8 billion into just 26 Indian tech startups since the start of 2014. However, due to India's cumbersome bureaucracy, many of these companies are based in the U.S., even though much of their workforce is located in India, where labor is much less expensive.

Planarian Regeneration Model Discovered by Artificial Intelligence
Tufts Now (06/04/15) Kim Thurler

Biologists at Tufts University have for the first time used an artificial intelligence (AI) system to reverse-engineer the regeneration mechanism of the planaria, a small worm that can regrow its body parts. The researchers say this achievement is the first comprehensive model of planarian regeneration, and also the first model of regeneration discovered by a non-human intelligence. Researchers Michael Levin and Daniel Lobo developed an algorithm that uses evolutionary computation to produce regulatory networks in an effort to evolve an accurate model that could predict the published results of previous laboratory experiments on the planaria. It took the algorithm 42 hours to return a regulatory network that correctly predicted the results of the 16 experiments they used as their data set. Levin says the study represents a successful application of robot science and demonstrates that such systems can do more than simply help scientists crunch massive datasets. He notes creating theories, such as the model generated by his and Lobo's algorithm, "is pretty much the most creative, intuitive aspect of the scientist's job," and the study suggests AI systems could be just as capable as human researchers of generating such theories.

The Robot That Learns Everything From Scratch
Gemini (06/02/15) Steinar Brandslet

Researchers at the Norwegian University of Science and Technology (NTNU) have developed [self.], a robotic system designed to learn entirely from sensory input with no pre-defined knowledge database, so its learning process will resemble that of a human child in early life. The robot analyzes sound through a system based on the human ear, and learns to recognize images using a digital model of how nerve cells in the brain interact with sensory impressions. The robot has been on public display, where visitors affected its learning. Having [self.] interact with a diverse audience enabled the researchers to study exactly how the robot learns. The robot gradually absorbed more and more information, and learned to filter certain input. After a period of time, the robot learned to connect words and pictures together in a more complex manner, autonomously connecting sounds with images. "We believe that the right way to reach for the 'holy grail' of [artificial intelligence] is to implement biologically inspired models in a machine, let it operate in a physical environment, and see if we can observe intelligent behavior," says NTNU researcher Axel Tidemann. He says the goal of the [self.] project is enabling the robot to learn through interacting with humans as well as possible.

Penn Engineers Show How 'Perfect' Materials Begin to Fail
Penn News (06/02/15) Evan Lerner

The cause of defect-free material breakage, which can cascade into failure, is described by a University of Pennsylvania/Max Planck Institute for Intelligent Systems project. The researchers stretched defect-free palladium nanowires and found the stretching force at which they failed to be unpredictable, occurring in a range of values where ambient temperature wielded a much stronger influence than expected. The thermal uncertainty in the failure limit implies the point where a failure-inducing defect first manifests is on the wire's surface, where atoms exhibit more liquid-like behavior and their higher mobility makes their reconfiguration into the start of a line defect more probable. "Our goal was to deduce the point where the first of the nanowire's atoms begin to shift out of their original positions and form a mobile defect," says University of Pennsylvania professor Daniel Gianola. The researchers mapped out failure points by stretching the nanowires at various temperatures, confirming thermal activation. "When you make these really small structures, they're often grown from the bottom up...and that can give you a much more pristine structure than if you were to take a big block of metal and whittle it down," Gianola notes. "In addition, the atoms on the surface comprise a much larger proportion of the total and can control the properties of the nanoscale material."

Meraculous: Deciphering the 'Book of Life' With Supercomputers
Berkeley Lab News Center (06/01/15) Linda Vu

A team of multi-institutional researchers has streamlined and accelerated genome assembly from a months-long process to minutes via the development of Meraculous, a tool created using algorithms, computational methods, and the Unified Parallel C (UPC) programming language. "Using the parallelized version of Meraculous, we can now assemble the entire human genome in about eight minutes using 15,360 computer processor cores," says University of California, Berkeley graduate student Evangelos Georganas. "With this tool, we estimate that the output from the world's biomedical sequencing capacity could be assembled using just a portion of [the National Energy Research Scientific Computing Center's] Edison supercomputer." To make efficient use of massively parallel systems, Georganas developed an algorithm for de novo assembly that exploits the one-sided communication and Partitioned Global Address Space (PGAS) capabilities of the UPC programming language. PGAS enables researchers to treat the physically independent memories of each supercomputer node as one address space, reducing the time and energy the supercomputer spends swapping information between nodes. Lawrence Berkeley National Laboratory Computational Research Division researcher Leonid Oliker says the new parallel algorithms enable the rapid performance of assembly calculations. Meraculous developer Jarrod Chapman thinks this milestone could make metagenome analysis by the tool possible.

Unlocking Nanofibers' Potential
MIT News (06/04/15) Larry Hardesty

A technique developed by Massachusetts Institute of Technology (MIT) researchers promises to make nanofiber production more affordable. The approach increases the rate of production fourfold while reducing energy consumption by more than 90 percent, according to MIT Microsystems Technology Laboratories researcher Luis Fernando Velasquez-Garcia. The team adapted the electrospinning technique of applying a voltage between a rotating drum covered by metal cones and a collector electrode on a smaller scale. The researchers used techniques common in the manufacture of microelectromechanical systems. They were able to pack 225 emitters, each several millimeters long, on a square chip about 35 millimeters on a side. The emitters' small size lowers the voltage required to drive them and enables more of them to be packed together, boosting the production rate. "We have an array of emitters that can be thought of as a dot-matrix printer, where you would be able to individually control each emitter to print deposits of nanofibers," Velasquez-Garcia says. Nanofibers have a wide range of potential applications, from solar cells to water filtration to fuel cells, but have been limited to a few niche industries due to high manufacturing costs.

Quest for Buried Knowledge Continues With New Computer Software Tool
University of Kentucky News (06/03/15) Kody Kiser; Amy Jones-Timoney; Whitney Harder

A new computer software tool will enable scholars to read the writings of ancient scrolls without having to unroll or open them. The Volume Cartographer was developed by University of Kentucky professor Brent Seales and colleagues. The tool is designed to map the surface of a scroll, enabling the user to see layers of pages. Each scroll page becomes an uneven three-dimensional page that can be pulled out. The user can texture a page to flatten it into a two-dimensional equivalent and then can scan for letters to see if words are on the page. "It's really about what we can enable scholars to do," says project manager Seth Parker. "We want to create a pipeline that we can actually give to historians, classicists, the people who want to study these texts, and enable them to unlock their own artifacts." The tool currently is being used to reveal text from a Herculaneum scroll carbonized by the eruption of Mount Vesuvius in 79 A.D. "We are now poised for discovery--discovery not just of new technical methods and software development--but of texts that we might somehow rescue," Seales says.

SDSC, UCSD Focus on Sustainable Computer Science Courses
UCSD News (CA) (05/27/15) Jan Zverina

The University of California, San Diego's San Diego Supercomputer Center (SDSC) has been awarded a three-year U.S. National Science Foundation grant to help three regional school districts create model "villages" for deploying and sustaining up-to-date computer science (CS) courses in their curriculum. "Workforce training must evolve with technology innovations to maintain a vibrant economy," says SDSC's Diane Baxter. "The slower pace of K-12 curriculum revision poses a significant systemic challenge to an innovation-driven U.S. economy." Addressing this challenge is the purpose of the Computer Science-Creating a Village for Educators (CS-CaVE) initiative, which investigates a model designed to roll out new, tech-driven curriculum content faster than current systemic cycles of educational reform. Its aim is broader regional and national communities of educators with peer and university support elements that can be sustained indefinitely within operational district and school budgets. CS-CaVE employs a nationally-proven curriculum, and supports and examines how the school districts integrate this course into an overarching K-12 approach for introducing computing into the pre-college curriculum. The project also seeks to solve a key challenge confronting the national CS10K project--integrating a compelling computer science course into public high schools via a new Advanced Placement CS Principles test.

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