Association for Computing Machinery
Welcome to the December 4, 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).


Data Storage on DNA Can Keep It Safe for Centuries
The New York Times (12/03/15) John Markoff

Two separate recent experiments demonstrated the possibilities of encoding information in synthetic DNA molecules. The potential benefits of the technology include vastly greater longevity than current data storage media. Researchers believe it will soon be possible to create new hybrid storage systems thanks to the falling costs of synthetic DNA generation and sequencing. A project between the University of Washington and Microsoft has combined a desktop DNA sequencer with equipment used to amplify DNA fragments by producing billions of duplicates. The researchers say the prototype data-archiving system could be used for the long-term storage of digitized movies and medical images. They and another research team at the University of Illinois report successful storage on and retrieval of specific files from DNA. The researchers note rapid, efficient amplification of specific DNA strands via polymerase chain reaction can ease information retrieval. The University of Washington/Microsoft group thinks DNA's immense archival capacity makes it better suited for data storage than rewriting. In partnership with Twist Bioscience, the team plans to expedite production of custom DNA strands by scaling down the reaction for creating synthetic DNA. Both research teams believe a DNA-based storage system could potentially hold all of the world's digital data in about nine liters of solution.
View Full Article - May Require Free Registration | Return to Headlines | Share Facebook  LinkedIn  Twitter 

Growing Push to Expose More Students to Computer Science
Associated Press (12/02/15) Phuong Le

There is a growing effort nationwide, in collaboration with technology leaders, nonprofits, and companies, to expose more public school children to computer science. Schools in New York City, San Francisco, and other cities have committed to offer computer science to students in all grade levels, and Chicago says computer science eventually will become a high school requirement. By 2020, 4.6 million of 9.2 million forecast science, technology, engineering, and math jobs will be in computing, according to the U.S. Bureau of Labor Statistics. A key challenge is finding enough well-prepared teachers. To this end, the U.S. National Science Foundation is seeking to have 10,000 well-trained computer science teachers in thousands of high schools. Interest in computing has surged in recent years. Nearly 49,000 students took the College Board Advanced Placement (AP) exam in computer science in 2015, a 25-percent increase from the previous year. A new AP computer science principles course will be introduced next fall. The exam for that class is designed to be multidisciplinary, featuring real-world applications and aimed at increasing representation among women and underrepresented minorities. "Colleges are saying, 'there's an incredible demand, we'd like to see more students become better prepared when they get here,'" says the College Board's Terry Redican.

UW Roboticists Learn to Teach Robots From Babies
UW Today (12/01/15) Jennifer Langston

Babies learn by watching and imitating what adults are doing, and robots can "learn" in much the same way, according to computer scientists and developmental psychologists at the University of Washington (UW). The researchers have demonstrated robots can learn by amassing data through exploration, watching humans do something and determining how to perform that task on their own. "You can look at this as a first step in building robots that can learn from humans in the same way that infants learn from humans," says UW professor Rajesh Rao. The team used research on how infants follow an adult's gaze to develop machine-learning algorithms that enable a robot to explore how its own actions result in different outcomes. The researchers' robot used this learned probabilistic model to infer what a human wants it to do and complete the task, and even "ask" for help if it is not certain that it can. The researchers tested the robotic model in a computer simulation experiment in which a robot learned to follow a human's gaze. In another experiment, an actual robot imitated a human moving toy food objects around a tabletop.

The Future of Intelligence: Cambridge University Launches New Center to Study AI and the Future of Humanity
University of Cambridge (12/03/15) Fred Lewsey

The Leverhulme Trust has awarded a 10-million-pound grant to the University of Cambridge to study the impact of artificial intelligence (AI) on humanity. Although many leading AI researchers believe human-level AI in machines will be the biggest event in human history and the achievement could be reached this century, they note the ramifications of the technology need further study. Cambridge will establish a new interdisciplinary research center to explore the opportunities and challenges of AI technology. The Leverhulme Center for the Future of Intelligence will bring together computer scientists, philosophers, social scientists, and others to examine the technical, practical, and philosophical questions surrounding AI. Cambridge will lead a collaborative effort, which includes the University of Oxford, Imperial College London, and the University of California, Berkeley. Cambridge developed its winning proposal with its Center for the Study of Existential Risk (CSER). "The center is intended to build on CSER's pioneering work on the risks posed by high-level AI and place those concerns in a broader context, looking at themes such as different kinds of intelligence, responsible development of technology, and issues surrounding autonomous weapons and drones," says CSER executive director Sean O hEigeartaigh.

Coalition Calls for Greater Focus on Computer Science in UC, Cal State Admissions
Los Angeles Times (12/02/15) Carla Rivera

The University of California (UC) and California State University (CSU) systems should focus more on computer science in admissions, according to California Lt. Gov. Gavin Newsom and a coalition of high-profile business and academic leaders. In a letter sent Wednesday to the Board of Admissions and Relations with Schools, Newsom, Facebook's Sheryl Sandberg, Yahoo co-founder Jerry Yang, and others called for new admissions requirements for mathematics that include computer science as an option. Many education advocates believe an understanding of computing is now essential for many jobs. The coalition says modifying the college admissions requirements would provide school districts with an incentive to offer more computer science classes. The board is a committee within the UC Academic Senate that oversees admissions of undergraduate students. Under the proposal, the Academic Senate would still decide the level of academic rigor for courses. "In terms of preparing students for college for UC and CSU, we can't overlook establishing a competency in math," says UC Academic Senate chairman J. Daniel Hare. The Oakland Unified School District's Claire Shorall says computer science is an increasingly popular elective among students, but the drive is not as strong as it could be if it were included in the math requirement.

The Computer That Can Rate Your Therapist
The Washington Post (12/02/15) Matt McFarland

A psychotherapist's empathy toward their patients can be accurately measured by an automated computer system using only audio recordings of sessions, according to researchers from the universities of Southern California, Utah, and Washington. They employed ratings that experts had given to therapists in a series of previous sessions to train the system to recognize trends in word usage associated with empathetic or non-empathetic ratings. The system then received new recordings of sessions and could effectively assess the degree of empathy the therapist was exhibiting. Earlier research uncovered other empathy metrics besides words and phrases, including vocal energy, head motion, and correlations in speaking rhythms between therapist and patient. The new system is cued to pick out word choices indicating either empathetic reflective listening, or less-empathetic instruction and probing. The researchers say wide deployment of such a system could give therapists immediate feedback on their progress with patients. They note the historical therapist-rating methodology involves the costly and time-consuming technique of having a human observer sit in and identify relevant behavior and languages.
View Full Article - May Require Free Registration | Return to Headlines | Share Facebook  LinkedIn  Twitter 

Great Innovative Idea--Python Tutor
CCC Blog (12/02/15) Helen Wright

University of Rochester professor Philip J. Guo has developed a Web-based tool called Python Tutor, which automatically draws diagrams to help learners create mental models when learning computer programming. Python Tutor started as a project to visualize code written in the Python language, but has since expanded to include six additional languages: Java, Ruby, JavaScript, TypeScript, C, and C++. More than 1.5 million people from more than 180 countries have used Python Tutor to collectively visualize more than 13 million pieces of code. Students have used it when taking online courses, studying digital textbooks, and in traditional K-12 and college classes. The tool's users come from all age groups, with one-sixth of users over 55 years of age. As computer programming becomes an even more vital skill across more fields in the future, the impact of the tool is expected to grow. Guo also has helped his students build three social-learning systems on top of Python Tutor; Codechella enables multiple people to join a single Python Tutor visualization session, while Codeopticon enables a single tutor to monitor dozens of learners coding in real time and offer assistance if needed. Codepourri enables a crowd of volunteer learners to collectively create coding tutorials by annotating Python Tutor visualizations.

When Your Boss Is an Uber Algorithm
Technology Review (12/01/15) Tom Simonite

Uber and Lyft's automated management systems establish new dynamics between workers and their bosses that should garner regulatory attention, according to researchers with the nonprofit Data & Society. Drivers contracted by Uber or Lyft are typically required to meet a human when they first sign up, and afterwards they interact with an automated management system mainly accessed by a mobile app. When a driver is logged in, the app allocates pickup requests from people nearby, and the system assigns feedback by tracking the proportion of pickups a driver accepts and averaging the rating passengers give their driver following a ride. Drivers can be penalized for not accepting enough rides or for low ratings. Moreover, they have an incentive to work at particular times, or in particular places, by surge pricing that provisionally elevates fares. "Uber says it wants to act as a neutral intermediary that connects supply and demand with an automated mechanism for finding the right price," notes Data & Society researcher Alex Rosenblat. "It's difficult to argue that you're a neutral platform if you're actively trying to manage supply and demand." Rosenblat says regulators should consider whether the way such automated management systems can manipulate employees' behavior could exceed ethical boundaries.

Researchers Enlist Gamers to Find Something Fishy About Mathematical Models
Uppsala University (12/02/15) Linda Koffmar

Uppsala University researchers recently implemented a Turing test in the form of an online game, with more than 1,700 players, to assess how good mathematical models were at reproducing the collective motion of real fish schools. The online game involved members of the public and a small group of experts being asked to differentiate between the collective movements of real fish schools and those simulated by a model. "By putting the game online, and through crowdsourcing this problem, the public have not only become engaged in science, they have also helped our research," says Uppsala University researcher James Herbert-Read. Although the statistical properties of the model matched those of the real data, both experts and members of the public were able to tell the difference between simulated and real fish. The researchers asked online players that answered all six questions correctly to give feedback on how they were able to differentiate between real schools and simulated ones. The players suggested the spatial organization of the groups and smoothness of the trajectories appeared different between the simulated and real schools. "Our results highlight that we can use ourselves as Mechanical Turks through 'citizen science' to improve and refine model fitting," Herbert-Read says.

Professor Is Designing Tools to Help Computers Sense Emotion
UT Dallas News Center (12/02/15) Chaz Lilly

University of Texas at Dallas professor Carlos Busso is using a $500,000 U.S. National Science Foundation Faculty Early Career Development Award to design speech-recognition tools that understand human emotion. "The shortcomings in current algorithms that recognize expressive behaviors during natural human interaction is the key barrier to using emotion-aware technology in real-life applications," Busso says. He wants to create new algorithms that recognize spontaneous behaviors from speech and capture the underlying externalization process of emotions in real-world conditions. The proposed models and algorithms could provide insights to explore theories in linguistic and paralinguistic human behaviors. "Several new scientific avenues can emerge that serve as truly innovative advancements that will impact applications in security and defense, the next generation of advanced user interfaces, health behavior informatics, and education," Busso says. He notes the first step is to identify speeches from open-access sources and social media to create a database of emotionally charged sentences to analyze. Busso then plans to use crowdsourcing techniques to enable trained ears to label the speech segments according to basic classifications of emotion, with the goal of reaching 100 hours of annotated emotional data.

What Makes Paris Look Like Paris? Let an Algorithm Tell You
Motherboard (11/30/15) Clinton Nguyen

Researchers from Carnegie Mellon University; the University of Illinois, Urbana-Champaign; the University of California, Berkeley, and INRIA sought to determine whether a machine could be trained to recognize a city better than humans. The team developed a program that took thousands of Google Street View photos and looked at randomly selected "patches" in the photos. Most of the patches were nondescript or frequently recurring, such as trees or windows, so the program focused on distinctive flourishes such as street signs, railings, or certain balcony styles, enabling it to considerably narrow down the results. During the trial, the program was able to pinpoint the distinguishing features of cities such as Paris, but U.S. cities, with their mishmash of architectural styles, were more difficult to identify. The researchers say the program could be used to create reference guides for cities that could leverage existing data instead of sending out people to scout locations. In addition, they say a computer could use it to generate a "Paris-like" set for a TV series. A paper describing the research appears in the December issue of Communications of the ACM.

The AMP Lab Stands Up to Big Data
Berkeley Research (11/30/15) Wallace Ravven

Researchers at the University of California, Berkeley's AMP Lab have developed several blockbuster, open source systems since its launch in 2011. Berkeley's Mike Franklin says the software system Mesos, which provides near-optimum access to needed resources in a cluster of computers, is used by Amazon, Twitter, Airbnb, Alibaba, Intel, and Google, among others. The lab's most influential invention is Spark, a software system that is used by hundreds of companies, government agencies, and researchers to process and analyze large volumes of data. Spark delivers data analysis to Autodesk, Novartis, and Microsoft, among many others. At a recent Spark Summit in New York, a managing director at Goldman Sachs described Spark as "the lingua franca of big data analysis." AMP stands for Algorithms, Machines, and People, and the lab's focus is to integrate this trinity in novel ways to improve data analysis. Franklin says the greatest challenges in data science will not be addressed by making machines do more of what they already have been doing. He says systems need to incorporate algorithms, machines, and crowds, and flexibly blend these resources. "We believe that will be the breakthrough," Franklin notes.

How Supercomputing Can Survive Beyond Moore's Law
IEEE Spectrum (12/02/15) Jeremy Hsu

A 10-exaflop or higher supercomputer will demand major changes in computing technologies and architectures, and Sandia National Laboratories' Erik DeBenedictis has collaborated with the IEEE Rebooting Computing initiative and the International Technology Roadmap for Semiconductors to help realize this vision. In an interview, DeBenedictis emphasizes the need for supercomputing to be extended beyond Moore's Law. "The pressures are going to be greater on supercomputing because they have driven up the usage and requirements for the logic portion--like floating-point logic--really high over the years," he says. "There's little slack and little room for improvement until they hit the physical limits." DeBenedictis says software developers must plan out products with much longer lifespans, as software tends to greatly outlast hardware. "If you can project what a supercomputer is going to look like in the second half of its lifespan for a piece of software, why not write that software for the computer it's going to run on for most of its lifespan, because we can predict it?" he says. DeBenedictis also sees a lack of communication between hardware and software developers. "Nobody is really thinking about how to code for the next generation of [multicore] processors," he notes.

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