Association for Computing Machinery
Welcome to the July 6, 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).


A Bug in fMRI Software Could Invalidate 15 Years of Brain Research
ScienceAlert (07/06/16)

The past 15 years of human brain research could be invalidated by a recently discovered bug in functional magnetic resonance imaging (fMRI) software, according to researchers at Linkoping University in Sweden. When researchers interpret data from an fMRI scanner, they are not studying the actual brain, but images of the brain divided into "voxels" interpreted by statistical software. Linkoping's Anders Eklund and colleagues obtained resting-state fMRI data from 499 healthy people sourced from databases worldwide, divided them into cohorts of 20, and measured them against each other to produce 3 million random comparisons. They then tested the three most popular fMRI software packages for fMRI analysis, which generated false-positive rates of up to 70 percent instead of the 5 percent the team originally expected. One of the bugs identified by the researchers had been in the system for 15 years, which is why up to 40,000 papers are potentially at risk of invalidation. The fact that the bug was not spotted for so long signals the ease with which such an error can occur, as researchers have lacked reliable techniques for validating fMRI results. Compounding the situation is the huge cost of fMRI scanning, as well as the prohibitively slow rate of result validation programs.

The Fully Self-Driving Car Is Still Years Away
The New York Times (07/01/16) Neal E. Boudette; John Markoff; Jack Ewing

The arrival of fully autonomous automobiles is complicated by the debate over how much self-supervision should be allowed, an issue reinforced by last week's disclosure the driver of a Tesla Model S was killed in a crash that occurred while the car was running in its Autopilot mode. Most automakers believe self-driving technology is not production-ready, with BMW projecting such vehicles will not be offered until 2021. Ford, Google, Volvo, and others are pushing toward fully autonomous cars that can operate safely without any human intervention. One step below that level of automation is semi-autonomous technology, which can drive the car for stretches of road under certain conditions but requires drivers to be ready to take over. "There's a huge inherent danger and it's well proven--the computer making a mistake and the driver not taking over quickly enough," says Alix Partners' Mark Wakefield. Mobileye co-founder Amnon Shashua, whose camera and sensor technology is used by Tesla, says self-driving technology is close, but still unready for deployment without human motorists remaining engaged. Shashua thinks the validation of the technology could be reasonably accomplished within five years, although the kind of vehicles Mobileye is working on may not necessarily be capable of full autonomy in complex urban settings.
View Full Article - May Require Free Registration | Return to Headlines | Share Facebook  LinkedIn  Twitter 

Lack of Role Models Keeps Women Out of Cybersecurity
Financial Times (07/06/16) Emma Jacobs

Ten percent of people working in the information security profession are women, the same percentage as two years ago, according to (ISC)2's 2015 Global Information Security Workforce Study. Cybersecurity professional Holly Rostill says this stagnant trend can be attributed to the lack of female role models in the field, which causes a cybersecurity-related field to seem unrealistic to many young women. "It's important that we attract the younger generation, to try and break the negative cycle that surrounds the industry for many women," Rostill says. In addition, young women need to be encouraged to get involved in science, technology, engineering, and mathematics, according to IOactive CEO Jennifer Stearns. She says young women need role models early in their careers who can highlight female professionals that have successfully broken through gender barriers. One issue surrounding the number of women in cybersecurity is recruitment. Cybersecurity is mainly discussed in terms of passwords and hacking, but there is a culture aspect that involves psychological engagement and organizational psychology, according to Deloitte organizational psychologist Cheryl Sims-Hancock. In addition, women need to feel fully qualified to apply for a job, whereas men are not as worried about being fully qualified. This view is backed up by (ISC)2's study, which found women in the cybersecurity field were more qualified than men.

New Translation Tool Will Help Facebook Master International Slang
Technology Review (07/01/16) Tom Simonite

A new Facebook feature rolled out last week uses automatic translation software to enable people to simultaneously post updates in multiple languages, and is providing the social network with data to help its translation software convert slang and other colloquialisms from one language to another. When a user composes an update, the new feature presents the option of clicking to create additional versions in different languages. The user can write those from scratch, or edit text that Facebook's translation software produces from their original post. Facebook software developer Necip Fazil Ayan says the innovation should help the program more effectively translate the language people often use when posting. Most translation systems have a tendency to use formal language because they are trained on collections of business, political, legal, and news documents translated by professionals. "A good quantity of parallel Facebook-post-style data would allow much, much better and more colloquial translations," notes Stanford University professor Christopher Manning. However, he cautions Facebook users may not themselves be adept at translating slang, and they might attempt to teach the system false translations.

How the Fifth Amendment Exposes a 'Serious' Flaw in the Turing Test (07/05/16) Emily Reynolds

Coventry University researchers Kevin Warwick and Huma Shah say the Turing Test for artificial intelligence has a "serious problem." According to them, if a machine were to take the Fifth Amendment (to the U.S. Constitution, which includes a prohibition against required self-incrimination) and remain silent throughout the test, it could potentially pass the test and be regarded as a thinking entity. The researchers used transcripts of machine-human communications in which machines remained silent--generally due to a machine or communications failure--and a human judge was unable to determine if they were interacting with a person or a machine. Warwick and Shah say a computer could be programmed with the personality of a middle-aged person who was sensitive about their age, and during a Turing Test could refuse to answer an age-related question, remain silent, and leave the judge uncertain as to whether they were talking to a human or a machine. "If an entity can pass the test by remaining silent, this cannot be seen as an indication it is a thinking entity, otherwise objects such as stones or rocks, which clearly do not think, could pass the test," Warwick says. The idea has received some criticism on Twitter.

Researchers Sue the Government Over Computer Hacking Law
Wired (06/29/16) Kim Zetter

Four academic researchers are suing the U.S. government over the federal Computer Fraud and Abuse Act (CFAA), which prevents them from conducting research to uncover instances of illegal discrimination by proprietary algorithms used by websites to determine what data to show visitors. "Big data enables behavioral targeting, meaning that websites can steer individuals toward different homes or credit offers or jobs--including based on their membership in a class protected by civil rights laws," the plaintiffs say. As a result, they say "[b]ehavioral targeting opens up vast potential for discrimination against marginalized communities." The plaintiffs claim a provision of the CFAA could be applied to criminally prosecute them for scraping publicly available data from these sites or creating anonymous user accounts on them, if the sites' terms of service ban such activity. With publisher First Look Media Works joining the lawsuit, the plaintiffs contend opening bogus profiles to pose as job and housing seekers is speech and expressive activity shielded under the U.S. Constitution's First Amendment. They also argue since websites can change their terms of service at any time without notifying visitors, this can criminalize any speech or activity on the site, which violates the Constitution's Fifth Amendment right to due process. Their issue is the threat of prosecution enables websites to stifle research that helps determine if those sites are violating the law.

Grade-School Students Teach a Robot to Help Themselves Learn Geometry
New York University (06/30/16)

A New York University researcher and colleagues from Arizona State University and Carleton University say they have developed educational technology that could better engage students. The team's tangible learning environment utilizes teachable agent framing coupled with a physical robotic agent. Robo-Tangible Activities for Geometry (rTAG) projects a Cartesian plane onto a white floor mat, upon which a LEGO robot named Quinn navigates. An iPod Touch mounted on top of the LEGO components displays Quinn's face and outputs its voice, through which it can give affective responses. The iPod also provides the entry point for interacting with Quinn. The final component is the mobile interface, which is another iPod Touch held by a student when interacting with the system. Students issue a command to Quinn by first touching the mounted iPod, which triggers a popup on the mobile interface, where they can choose from a variety of actions, including move units, turns, and plot points. A paper on the rTAG tangible learning environment was presented at the ACM CHI 2016 conference in Santa Barbara, CA in May.

This Robot Follows You Around and Blasts You With Air Conditioning
The Washington Post (06/30/16) Darryl Fears

More efficient air conditioning via a mobile robot called a Roving Comforter (RoCo) is a goal of researchers at the University of Maryland's (UMD) Center for Environmental Energy Engineering led by Reinhard Radermacher. The researchers are participating in a three-year competition hosted by the U.S. Department of Energy's (DOE) Delivering Efficient Local Thermal Amenities (DELTA) program. RoCo follows its owners like a self-propelled vacuum cleaner and provides sufficient comfort to permit residences and businesses to adjust the thermostat up to four degrees. The UMD researchers received a $2.5-million DOE grant to develop a robot that would power on by itself, monitor motion, generate cool air, and store the heat produced in the process. In response to a signal via wearable technology or a smartphone app, RoCo adjusts its height to custom-cool its user. Radermacher says RoCo blows cool air on a person's head or chest when it comes within three feet of him. "It will allow people to be comfortable in their environment without playing with the thermostat," says UMD's Daniel Dalgo. Jennifer Gerbi with the DOE's Advanced Research Projects Agency-Energy thinks RoCo would be very appealing to warehouse operators seeking to cut energy costs in large rooms with a handful of employees, but she is less certain such robots could find use in crowded offices.
View Full Article - May Require Free Registration | Return to Headlines | Share Facebook  LinkedIn  Twitter 

Tackling Intractable Computing Problems
National Science Foundation (06/29/16) Aaron Dubrow

Princeton University researchers used a $10-million U.S. National Science Foundation (NSF) Expeditions in Computing award to find the sources of intractability, to circumvent intractability when possible using approximations and other means, and to understand the implications of intractability for other areas. Intractability means it might not always be possible to find accurate, efficient algorithms for some important problems. The researchers made several theoretical advances, including a new understanding of the Unique Games Conjecture. They found there is an algorithm for solving Unique Games that is better: one faster than exponential-time, but still not as fast as polynomial-time. "Some of the important leaps, for example in the understanding of efficient approximation, public-key cryptography, arithmetic complexity, and the role of intractability in other sciences, seem to happen mainly due to the collaborative environment of the Intractability Center, supported by the NSF Expedition," says Avi Wigderson, a researcher at Princeton's Institute for Advanced Study. In addition, the project served as a training ground for theoretical computer scientists. "The team also innovated in attracting new talent to theory via new summer programs for undergraduates and high school students, and a highly successful workshop where established women in theory reached out to undergraduate and entering graduate student women," says NSF's Tracey Kimbrel.

New Framework Uses Patterns to Predict Terrorist Behavior
Binghamton University (06/28/16)

Binghamton University researchers have developed a framework that can predict future terrorist attacks by recognizing patterns in past attacks. The researchers used data on more than 140,000 terrorist attacks between 1970 and 2014 to develop the framework, which calculates the relationship between specific features of terrorist attacks, such as attack time and weapon type. The framework identifies the characteristics of future terrorist attacks by analyzing the relationship between past attacks. The proposed method was able to successfully predict most of the characteristics of attacks with more than 90-percent accuracy, according to the researchers. The results support previous findings that terrorists tend to mimic the behavior of other terrorist groups and learn from their mistakes and successes. The framework works to define which metrics are important, according to Binghamton University researcher Salih Tutun. He says policymakers could use this approach for time-sensitive understanding and detection of terrorist activity, which can enable precautions to protect against future attacks. "Predicting terrorist events is a dream, but protecting some area by using patterns is a reality," Tutun says. "If you know the patterns, you can reduce the risks. It's not about predicting, it's about understanding."

Google Seeks to Spur Kids' Interest in Coding With Project Bloks
eWeek (06/28/16) Jaikumar Vijayan

Google is working to get kids started on programming at a very young age by collaborating with design firm IDEO and a Stanford University researcher on Project Bloks, an initiative that builds on research in tangible programming in which children learn basic programming concepts by manipulating physical objects such as wooden blocks. Google says Project Bloks features an open hardware platform that will give designers, developers, educators, and others a way to build "physical coding experiences" for children. As a first step in this direction, Google has built a working prototype of a system for tangible programming that consists of three components--a "Brain Board," "Base Boards," and programmable "Pucks." Each puck can be programmed with a different function and be placed on the Base Board, which then reads the instructions on the puck via a capacitive sensor. When connected to the Base Boards, the Brain Board can send the instructions via Bluetooth or Wi-Fi to connected devices. Robots, toys, and other connected devices then execute the instructions. "As a whole, the Project Bloks system can take on different form factors and be made out of different materials," say Google's Steve Vranakis and Jayme Goldstein. "This means developers have the flexibility to create diverse experiences that can help kids develop computational thinking."

Stanford Research May Lead to More Durable Electronic Devices Such as Cellphones
Stanford News (06/27/16) Andrew Myers

Stanford University researchers have found the glassy materials in transistors respond very differently to compression than they do to the tension of bending and stretching. The glassy materials "are actually stiffer when compressed than when stretched, and we can use this knowledge to design more durable chips and devices," says Stanford professor Reinhold Dauskardt. When active, electronic devices heat up and the components expand; when not in use, the components cool down and contract. The materials' response to this expansion and contraction is inherently related to the interaction within the network of atoms or groups of atoms that do not fully bond during production. In compression, these atoms strongly repel each other to make the network stiffer. In tension, their failure to bond causes the same atoms to interact less, making the materials less stiff and, consequently, more expansive than expected when they heat up. "If you could get rid of all of the unbonded terminal groups and create an absolutely flawless material, you would not see these asymmetries, but we can't, so we have to understand and accommodate this knowledge in design," Dauskardt says. Thanks to the Stanford research, materials scientists will have to integrate the results into their mathematical algorithms, which currently assume the stiffness is symmetrical.

'Socially-Cooperative' Cars Are Part of the Future of Driverless Vehicles, Says CMU Professor
TechRepublic (06/29/16) Hope Reese

In an interview, Carnegie Mellon University (CMU) professor John Dolan says his team is investigating "socially-cooperative driving" as an enhancement to self-driving vehicles, using probability to judge driver intent and communicate motorists' intent to other drivers so interaction with other vehicles is safer and more natural. "Having humans involved is desirable because of the fact that the technology's not sufficiently matured to avoid it," Dolan says. He says CMU's explorations include retrofitted vehicles with two people in the car--a safety driver who can switch the car from autonomous to manual when necessary, and a developer testing and tweaking the software as needed. "If there's a situation we can't handle because it's too complex, we intervene in order not to cause any problems," Dolan notes. He cites challenges to self-driving cars such as inhibited global-positioning system accuracy in urban settings and inclement weather. Methods Dolan's team is pursuing to read stoplights accurately include vision/camera systems and communication with the light itself. Areas needing significant progress include developing a lower-cost, highly precise localization system, and perception algorithms, according to Dolan. He notes many automakers are approaching fully autonomous vehicles using advanced driver-assistance systems, and he says deep learning "can have an increasing impact on driving, improve some components of it."

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